ibaan Diagnostic Analyzer 1.0 User's Guide for Diagnostic Analyzer 1.0

Transcription

1 ibaan Diagnostic Analyzer 1.0 User's Guide for Diagnostic Analyzer 1.0

2 A publication of: Baan Development B.V. P.O.Box AC Barneveld The Netherlands Printed in the Netherlands Baan Development B.V All rights reserved. The information in this document is subject to change without notice. No part of this document may be reproduced, stored or transmitted in any form or by any means, electronic or mechanical, for any purpose, without the express written permission of Baan Development B.V. Baan Development B.V. assumes no liability for any damages incurred, directly or indirectly, from any errors, omissions or discrepancies between the software and the information contained in this document. Document Information Code: U7867A US Group: User Documentation Edition: A Date: May, 2002

3 Table of contents 1 Introduction About ibaan Diagnostic Analyzer What questions vs. Why questions 1-1 Structured search processes 1-2 Diagnostic modeler architecture and components 1-3 The Tools Diagnostic Analyzer Modeler Diagnostic Analyzer Reporter 1-8 Diagnostic Analyzer Tree Analyzer Sample diagnostic models To create diagnostic models To create a diagnostic model Procedure summary 2-1 Procedure description The basic features of the Diagnostic Analyzer Modeler The Modeler window 2-4 The toolbar 2-5 The pull down menus 2-9 The context menu General Model Properties 3-1 The Edit Model dialog 3-1 SQL Server Connections 3-4 Usage of Model variables To define intermediate nodes 4-1 Procedure overview 4-2 To define the intermediate node s properties 4-3 Usage of SQL Queries 4-7 To define SQL queries 4-7 To view SQL Query syntax To define conclusive nodes Procedure overview To define the conclusive node s properties To define branches 6-1 Procedure overview 6-2 i

4 Table of contents To define the branch s properties To define queries To start the Query Builder To select tables 7-3 To define joins (optional) 7-4 To select table fields 7-5 To define expressions (optional) To define conditions (optional) To define subqueries 7-8 To preview the query results To run a diagnostic model To define the default report settings for running a model To run a model from the Modeler 8-5 To run a model through the Reporter 8-5 The Diagnostic Analyzer Reporter dialog 8-6 Execute a model through the Extraction Job Runner Output Conclusion sheet 8-10 Intermediate results file 8-10 Conclusion table To analyze the Result tree 9-1 Report settings To analyze the Result tree Tree Analyzer structure 9-3 The Result Set Pane The Control Panel Tutorial 10-1 The model 10-1 Steps to build the model Some overall model properties The first intermediate node 10-4 Adding additional Model Variables 10-7 The second intermediate node 10-8 Connecting the first two nodes and editing the Branch The third intermediate node Connecting the second and the third node The fourth Intermediate node The remaining nodes Prepare for execution ii

5 About this document This document is a manual for users who want to develop and run structured search models through the ibaan Diagnostic Analyzer. Chapter 1, Introduction, introduces the ibaan Diagnostic Analyzer. Chapter 2, To create diagnostic models gives an overview of the Diagnostic Modeler and the procedure to create diagnostic models. Chapter 3, General Model Properties, describes how to define the basic properties for a diagnostic model. Chapter 4, To define intermediate nodes, describes how to create intermediate nodes in a diagnostic model. Chapter 5, To define conclusive nodes, describes how to create conclusive nodes in a diagnostic model. Chapter 6, To define branches, describes how to connect nodes through branches. Chapter 7, To define queries, describes how to create SQL queries, in intermediate nodes, through the Query Builder. Chapter 8, To run a diagnostic model, describes how to run a diagnostic model from the Modeler or through the Reporter. Chapter 9, To analyze the Result tree, describes how to investigate the search results through the Tree Analyzer. Chapter 10, Tutorial, contains systematic instructions for creating a simple diagnostic model, based on the Northwind sample database. Additional documents ibaan Diagnostic Analyzer Installation Guide for Diagnostic Analyzer 1.0 (U7868A US) ibaan Decision Manager 2.2 Administrator s Guide (U7671A US) iii

6 About this document iv

7 1 Introduction About ibaan Diagnostic Analyzer ibaan Diagnostic Analyzer is a tool for structured searching, as opposed to ad hoc querying or unstructured searching, for the causes of operational problems. It aids decision-making and maximizes the benefit of the data in a Data Warehouse. Business analysts can use it to investigate operational problems of any kind, as long as these problems are measurable and quantifiable using the data accumulated in the transactional (ERP) system. ibaan Diagnostic Analyzer is particularly powerful when used on top of a Data Warehouse, in which data is structured in star schemas of fact tables and dimension tables. Some distinguishing features of ibaan Diagnostic Analyzer are: Analysis is done through predefined search models reflecting the business processes that played a (possible) role in the operational problem. Executing a search model renders a weighted list of to-the-point conclusions that can directly aid the decision making process. The product comprises a fully integrated set of tools for building search models, defining and testing queries, running search models, and viewing search results. ibaan Diagnostic Analyzer requires Microsoft NT4 or Windows 2000, Microsoft SQL Server (version 7.0 or higher), and Microsoft Internet Explorer (version 5 or higher). What questions vs. Why questions Traditional BI tools (OLAP analysis, business reporting) focus on answering What questions: What was the delivery reliability for this item in the last month? What was the turnover by sales representative this year? What was our profitability this year and how does this compare with previous years? 1-1

8 Introduction They produce performance indicators that could be rated positive or negative. In both cases, we want to find the causes and get an answer to the Why question: Why was the delivery reliability for this item in the last month bad? Why was the turnover of this sales representative below average? Why was the profitability in this year so much higher than previous years? Answering these questions can be highly complicated. Take the first question. To find the bottleneck, we must check the entire upstream logistic process. ibaan Diagnostic Analyzer uses structured search processes to answer this type of Why questions. Structured search processes For many areas in the domains of Logistics, Finance, Service, the underlying data-generating process is well known because it is a well-defined business process (even being controlled by an ERP system). Structured search processes exploit this knowledge and thus are expected to provide results that data mining techniques cannot offer. In a search process, the analysts start at the point where the problem has become visible. They collect and analyze information to identify the direct causes that explain the visible problem. Because they know the underlying business process, it is relatively straightforward to identify all potential direct causes. These causes now become new visible problems themselves, for which the analysts again must identify the potential direct causes. They must repeat this action until, for all chains of causes, the boundary of the system under consideration has been reached, that is where his knowledge ends and the basic information is gathered to answer the initial question. They now have broken down his problem into a number of subproblems and have changed their original why questions into a (large) number of what questions. Examples EXAMPLE 1 In a make to stock environment, in order to answer the question Why is item I delivered unreliably last month?, you can start the analysis with the following What question: Which warehouse delivered the item unreliably last month?. Suppose it turns out that this is warehouse W, the next topic to analyze could be What was the stock level and inflow of item I in warehouse W last month? 1-2

9 Introduction EXAMPLE 2 Alternatively, in a make-to-order environment, in order to answer the question Why is item I delivered unreliably last month?, the first topic to analyze could be What was the average time available (promised date - order date) to fulfill the orders for item I last month?. If this is too low, the company has been suffering from rush orders, so a next logical step is to analyze the performance of the sales order generation process. Otherwise, the manufacturing process is unreliable, so we must divert attention in that direction. This example shows two potential direct causes for the problem of unreliable deliveries: (1) Problems in the sales order generation process. (2) An unreliable manufacturing process. In the Diagnostic Analyzer, you can model and run this type of structured search processes: you can develop and execute diagnostic analysis models (diagnostic models). Diagnostic modeler architecture and components The Diagnostic Analyzer architecture consists of three applications (Diagnostic Modeler, Diagnostic Reporter, and Tree Analyzer) and various related components. See the following figure: Figure 1 Diagnostic Analyzer architecture 1-3

10 Introduction The various components in this architecture have the following roles: Diagnostic Analyzer components Component Role Search Model Diagnostic Modeler Report Definition Diagnostic Reporter Diagnostic Engine Result Tree Data source Conclusion Sheet Tree Analyzer OLAP cube OLAP Client A file containing the complete description of one search model. An application with which you can build and maintain search models. A file describing a concrete problem to be evaluated. It specifies a search model, the start node within the model, and values for the required context variables of this start node. Furthermore, a report definition can specify several parameters that influence the behavior of the search algorithm. An application with which you can create a Report Definition. An application component that runs the search algorithm. The Result tree of the algorithm. It can be stored in a file. It is also available In-RAM when it is being calculated / analyzed. The data provider for the analyses done with the model. The search results of a diagnostic model can be written to a conclusion table in this data source, for reporting needs and OLAP analysis. The most simple output format: a list of conclusions in order of decreasing weight. This is presented through an HTML file. An application used to analyze the Result tree in full details (all intermediate steps that led to the conclusions). A multi-dimensional storage of conclusion data suited for analyzing aggregated conclusions. Cubes are not delivered with the ibaan Diagnostic Analyzer. They are put in the overview because they can be very helpful for the analysis of conclusions (for example, aggregate the weight of all conclusions that relate to a certain customer, aggregate the weight of all conclusions of a certain type, and so on). An analysis tool for analyzing the OLAP cube. (Crystal Analysis is the preferred OLAP client for ibaanbis solutions. This tool is not provided as part of ibaan Diagnostic Analyzer.) 1-4

11 Introduction The Tools The Diagnostic Analyzer consists of the following tools: Diagnostic Analyzer Modeler. Diagnostic Analyzer Reporter. Diagnostic Analyzer Tree Analyzer. In addition, a number of sample diagnostic models is delivered with the Diagnostic Analyzer software. Diagnostic Analyzer Modeler The Diagnostic Modeler is a graphical tool to build search models. These search models are named Diagnostic Analysis Models or Diagnostic Models. A Diagnostic Model contains all possible pathways that can be followed when analyzing the data in order to find the causes for an operational problem. In the Diagnostic Analyzer Modeler, a search model is visualized as a graph with nodes and branches (arrows). Two types of nodes exist: intermediate nodes. An intermediate node specifies the analysis that must be done to solve one step in the search process. More specifically, an intermediate node contains a query. The analysis in an intermediate node is mostly done in a specific context, for example for a particular item or for a particular customer. In each intermediate node, you can define a set of context variables, which determines the context for the node. Context variables in intermediate nodes are often used as parameters in the underlying SQL queries. For example: an SQL query uses the <[AnalysisYear]> parameter. The value for this parameter can be the result of an SQL query in a previous intermediate node or can be user input. Conclusive nodes. A conclusive node represents a conclusion that can be drawn on the basis of the analysis in an intermediate node. A conclusion is often specific for a combination of context variables, for example Supplier and Year: Supplier X had a bad delivery performance in year Y. Nodes are connected by branches: branches determine the direction in the search process. They also determine the choices (such as can we already draw a conclusion or should we continue our analysis ) that can be made. 1-5

12 Introduction The following figure shows an example of a model created with the ibaan Diagnostic Analyzer. The upper six shapes in the left part of the model are intermediate nodes; the other shapes are conclusive nodes. Figure 2 Sample diagnostic model 1-6

13 Introduction NOTE This sample model is based on the Northwind database, which is delivered with Microsoft SQL Server. It tries to find an explanation for the below-average turnover of employees in a particular calendar year. The conclusions generated by a model run will be employee specific. The employee with the largest gap between target (or average) and his / her personal turnover (order value) will be on top of the conclusion list. An octagon shape indicates a potential start node. See the first and third node in the figure. You can change the appearance (colors, shapes, fonts, and so on) of nodes and branches in the Graphical options dialog. To start this dialog, select Graphical Options from the View menu or from the context menu. The analysis in the sample diagnostic model comprises the following steps: 1 Av. Discount and freight. In this first node (one of the two possible start nodes), two overall averages are calculated: The average discount percentage for all employees. The average amount of freight costs per order for all employees. These two numbers are used as a standard for classifying individual employees. 2 Average Order Value. In the second node, the average turnover is calculated for the calendar year of interest. 3 Find bad performers. In this node, the turnover of each individual employee is compared with this average turnover. The sales data for those employees that perform below average will be further analyzed in order to find the possible cause. This third node can also be used as an alternative start node. In that case, the user must specify the average turnover (or set a target), the average freight costs (or target), and the average discount percentage (or target). 4 Employee experience. In this node the experience of the employee is checked. If the employee is inexperienced (< 30 month at service), then we jump to the conclusion 'inexperienced employee'. Otherwise, we continue the analysis. 5 Average Discount. A possible explanation for the below-average (or belowtarget) individual turnover could be the high discounts that are given away by the employee. 6 Average Freight. If this is not the case, we check whether the freight costs for the orders closed by the employee were excessively high (that is, above the average). 7 No explanation. If all these explanations fail, then we have to conclude that there is no measurable explanation for the disappointing performance of the employee. 1-7

14 Introduction NOTE For details about the Diagnostic Analyzer Modeler, such as panes and menu commands, refer to the next chapter. Diagnostic Analyzer Reporter This is a tool for executing search models built with the Modeler. You can specify various parameters, such as: The diagnostic model to run. Name and location of the conclusion sheet. The initial value for the context variables of the start node. See the following figure: Figure 3 Diagnostic Analyzer Reporter 1-8

15 Introduction Executing a model renders a Conclusion Sheet. See the following figure for an example. Figure 4 Sample conclusion sheet NOTE The example in the screenshot shows the outcome of a model based on the Northwind database. The operational problem to be answered by this model is: Why do some employees have a below-average turnover? You can start the Reporter from the ibaan Diagnostic Analyzer menu, and through the ibaan Decision Manager Extraction Job Runner. The latter enables you to schedule an automated analysis, for example directly after each data warehouse upload. For details about the Diagnostic Analyzer Reporter, refer to chapter 8, To run a diagnostic model. For details about the ibaan Decision Manager Extraction Job Runner, refer to ibaan Decision Manager 2.2 Administrator s Guide (U7671A US) Diagnostic Analyzer Tree Analyzer The Tree Analyzer can be used to make a detailed investigation of the Result tree. This is the tree of result nodes produced by a model run. A result node is created each time a model node is visited, that is: Each time the analysis defined in a node is actually carried out with specific values for the context variables (intermediate result node). Or: 1-9

16 Introduction Each time an actual conclusion can be drawn (conclusive result node). See the following figure for an example. Figure 5 Tree Analyzer with sample Result tree NOTE For each result node, you can view some details of the result set and technical information, such as the context variables and the SQL query that was executed to obtain the result. For details about the Diagnostic Analyzer Tree Analyzer, refer to chapter 9, To analyze the Result tree. Sample diagnostic models The following sample diagnostic models are delivered with the software: Northwind. Diagnostic model for employee turnover analysis, based on the Northwind database. See the screenshot in the Diagnostic Analyzer Modeler section in this chapter. The Northwind database is a sample database that is delivered with Microsoft SQL Server. Alternative Suppliers. Diagnostic model for supplier analysis, based on a Baan IV data warehouse. This model identifies bad suppliers and tries to find alternative suppliers for the items they deliver. 1-10

17 2 To create diagnostic models This chapter covers the following topics: The procedure to create a diagnostic model. The basic features of the Diagnostic Analyzer. To create a diagnostic model Procedure summary The procedure to create and run a diagnostic model consists of the following steps: 1 Add a new model. 2 Enter basic model properties. 3 Design the model, that is: define intermediate nodes, conclusive nodes, and branches. 4 Enter default report settings. 5 Run the model. Refer to the next section for more details about these steps. Procedure description To create and run a diagnostic model, start the Diagnostic Analyzer Modeler from the ibaan Diagnostic Analyzer menu and take the following steps: 1 Add a new model: select New Model from the toolbar or select New from the File menu. This will start the Edit Model dialog. 2 Enter basic model properties in the Edit Model dialog: Enter the name and descriptive information (metadata) for the new model. Link the model to the SQL Server database that contains the tables that you want to analyze, for example your ibaan Decision Manager Data Warehouse database. 2-1

18 To create diagnostic models EXAMPLE Declare the variables that will be used in the nodes in your diagnostic model. If required, you can add extra model variables later, when you design the model. Refer to chapter 3 General Model Properties for details about the Edit Model dialog. 3 Design the following model: define intermediate nodes, conclusive nodes, and branches. intermediate nodes. In an intermediate node, you must define an SQL query to solve one step in the search process. The analysis in an intermediate node is mostly done in a specific context, for example, for a particular item or for a particular customer. In each intermediate node, you can define a set of context variables, which determines the context for the node. Context variables in intermediate nodes are often used as parameters in the underlying SQL queries. For example: an SQL query uses the <[AnalysisYear]> parameter. The value for this parameter can be the result of an SQL query in a previous intermediate node or can be user input. conclusive nodes. A conclusive node represents a conclusion that can be drawn on the basis of the analysis in an intermediate node. A conclusion is often specific for a combination of context-variables, for example, Supplier and Year: Supplier X had a bad delivery performance in year Y. Branches. Branches (arrows) are used to connect nodes: they determine the direction in the search process. You can define multiple branches to link a single intermediate node to various subsequent intermediate nodes and/or conclusive nodes. If multiple branches start in the same intermediate node, you can define a condition for each branch. A branch condition determines when the branch will be followed. See the following example: The Check Discount intermediate node calculates the discount ratio per sales representative. This node is connected to: A Discount Too High conclusive node. The corresponding branch is followed if the discount ratio is greater than 1. The conclusion Employee gives too high discounts is drawn. A Check Freight Costs intermediate node that calculates the freight cost ratio per sales representative. The corresponding branch is followed if the discount ratio is less than or equal to

19 To create diagnostic models IMPORTANT For details about nodes and branches, refer to chapter 4 To define intermediate nodes, chapter 5 To define conclusive nodes, and chapter 6 To define Branches. 4 Enter the default report settings. You can specify various parameters for running the model, such as: Name and location of the conclusion sheet. The initial value for the context variables of the start node. Maximum number of records to process in an intermediate result node. Name and location for the intermediate Results file / Result tree (optional). You can skip this step, if your model s current start node has no context variables. 5 Run the model. You can run your model through the Execute the model command in the Modeler, or through the Diagnostic Analyzer Reporter. If you specified an Intermediate Results file, you can use the Tree Analyzer to investigate the tree of result nodes, produced by the model run, in detail. For details about the Diagnostic Analyzer Reporter, refer to chapter 8, To run a diagnostic model. For details about the Diagnostic Analyzer Tree Analyzer, refer to chapter 9, To analyze the Result tree. Save your model after every important change. The basic features of the Diagnostic Analyzer Modeler This section covers the following topics: The Modeler window. The toolbar. The commands in the pull down menus. The commands in the context menu. 2-3

20 To create diagnostic models The Modeler window The following figure shows the main window of the Diagnostic Analyzer Modeler. Figure 6 Diagnostic Analyzer Modeler main window The Modeler is an MDI application. More models can be edited at the same time. Nodes can be copied (one by one) both within and between models. The main window consists of three panes: A drawing pane to sketch the model. Here, you can add, move or delete intermediate nodes, conclusive nodes and branches. If you double-click a component, a dialog where you can edit the component s properties is started. A Notes and Warnings pane: this lists reminders and omissions. Warnings are blocking the execution of the model. Notes are just for information. An Object Detail pane: this pane consists of two parts: The upper part shows the model layout. Here you can see the size and position of the displayed part of the model, compared to the model and the drawing area as a whole. 2-4

21 To create diagnostic models The lower part shows important details about the node or branch that is selected in the drawing area. For example: the context variables and the variables and conditions of the SQL query of the selected intermediate node are displayed. Two main types of nodes exist: Intermediate nodes (blue circle on the toolbar). Here, an analysis is defined (SQL query fired). The Start Node is a subtype: this is an intermediate node where an analysis can start. Conclusive nodes: here a conclusion is defined. You can connect an intermediate node to another intermediate node or to a conclusive node: first click the origin node, then the destination node (a dashed helper arrow will appear). Subsequently select the Connect Nodes by inserting a Branch toolbar button. To edit nodes and branches, you can Double-click the desired node or branch. Or: Select the desired node or branch and press Enter or F2. Or: Right-click the desired node or branch and select Edit selected object from the context menu. Or: Select the desired node or branch and select Edit in the Object Detail pane. NOTE You can select only one node or branch at a time. You can move nodes and branches with the mouse. Fine movements of nodes can be achieved with the ^, V, >, and < keys. You can segment branches: if you select a branch, segmentation tags appear. You can move these to the desired position. If you spoiled the branch (too many segments), you can straighten it again through a context menu command. The toolbar The following figure shows the Modeler s toolbar. Figure 7 Toolbar in Diagnostic Analyzer Modeler window The toolbar contains the following command buttons: New Model. To add a new diagnostic model. 2-5

22 To create diagnostic models Open Existing Model. Starts the Open Diagnostic Analysis Model dialog. In this dialog, you can browse to the desired windows directory, and open an existing model file (.dam) file. See the following figure. Figure 8 To open an existing model Save Model. To save the changes made to a model. The first time you save a model, the Save Model dialog is started. In this dialog, you can browse to the desired windows directory, and save your model in a model file (.dam) file. Show Model as HTML. Displays an HTML page with all meta data that is available in the model. This command is very helpful to get insight in the structure of a diagnostic model. The meta data that is displayed includes: Title and description of the model. A list of all variables used in the model. A list of all nodes used in the model Detailed information for each node: title, description, context variables, branches, and the syntax of the SQL query that is defined in the node. 2-6

23 To create diagnostic models See the following figure for an example. Figure 9 HTML page with metadata NOTE Delete selected Object. To delete the object you selected in the drawing pane. Cut Selected Node. To cut the intermediate or conclusive node you selected in the drawing pane. The node (not only the node s shape, but also the underlying settings, such as context variables and SQL syntax) is moved to the Modeler s internal clipboard and can be pasted in another model. You can open an extra model through the Open Existing Model button. You can use the Windows menu to switch from the current model to another opened model. Copy Selected Node. To copy the selected node and all underlying settings to the Modeler s internal clipboard. Paste Node. To paste a node and all underlying settings from the Modeler s internal clipboard into the current diagnostic model. 2-7

24 To create diagnostic models NOTE NOTE Edit Model Description, Variables, and Connection. Starts the Edit Model dialog. In this dialog, you can Enter the name and descriptive information (metadata) for the new model. Link the model to the SQL Server database that contains the tables that you want to analyze, for example your ibaan Decision Manager Data Warehouse database. Declare the variables that will be used in the nodes in your diagnostic model. Refer to chapter 3 General Model Properties for details about the Edit Model dialog. Edit default report settings. Starts the Default Report Settings dialog. Here you can specify default report settings for running the model and reporting the results, such as: Name and location of the conclusion sheet. The initial value for the context variables of the start node. Maximum number of records to process. Name and location for the intermediate results file (optional). For details about the Default Report Settings dialog, refer to chapter 8, To run a diagnostic model. Add intermediate node. Adds an intermediate node in the drawing pane. Subsequently, you can drag the new node to any position. To define/edit the node s properties, such as context variables and SQL Query, simply doubleclick the node. Refer to chapter 4, To define intermediate nodes for details. Add conclusive node. Adds a conclusive node in the drawing pane. Subsequently, you can drag the new node to any position. To define/edit the node s properties, such as context variables and conclusion text, simply double-click the node. Refer to chapter 5, To define conclusive nodes for details. Connect Nodes by inserting a Branch. Adds a branch between two nodes. To define a branch, select the first node, subsequently select the second node and finally select this button. To define/edit the branch s properties, such as branch condition and context mapping, simply double-click the node. Refer to chapter 6, To define branches for details. Execute The Model. Runs the model. Uses the default report settings for running, as specified in the Default Report Settings dialog. The results are displayed in a conclusion sheet. See the following figure for an example. For details about the execution of search models, refer to chapter 8, To run a diagnostic model. 2-8

25 To create diagnostic models Zoom. If you click the arrow next to the Zoom box, you can select the desired magnification from the list. Figure 10 Sample conclusion sheet The pull down menus You can select commands from the following pull down menus: File. Edit. View. Insert. Window. Help. File menu The File menu contains the following commands: New. To add a new model Open. To open an existing model. See the description of the Open Existing Model command in the The toolbar section in this chapter. Close. To close the current model. You are prompted to save the changes you made. 2-9

26 To create diagnostic models Save. To save the model. Save As. To save a model under a different name. SQL Server Connections. Starts a dialog, where you can add a connection to an SQL Server database, for example to your ibaan Decision Manager data warehouse database. You can define various SQL Server Connections to provide access to multiple databases. Each diagnostic model uses one SQL Server Connection to read data from the corresponding database. When you create a new model, you can select the desired SQL Server Connection from the list. To create a connection, you must enter the server name, login settings and the database name. See the following figure. Figure 11 To add a SQL Server Connection Run Model. To execute the model. Refer to the description in the The toolbar section in this chapter. Print Model Diagram. Prints the model, as it is displayed in the drawing pane. Show model as html. Displays an HTML page with all meta data that is available in the model. For details, refer to the description in the The toolbar section in this chapter. Exit. To close all models and to quit the Modeler. You are prompted to save the changes you made. 2-10

27 To create diagnostic models Edit menu The Edit menu contains the following commands: Model. Starts the Edit Model dialog, where you can enter basic model properties. Refer to chapter 3 General Model Properties for details Default report settings. Starts the Default Report Settings dialog. Here you can specify default report settings for running the model. For details, refer to chapter 8, To run a diagnostic model. Select (next) Object. Selects the next node or branch in the drawing pane. Edit Selected Object. Starts a dialog, where you can edit the properties of the selected node or branch. Delete, Cut, Copy, Paste. See the description of these commands in the The toolbar section in this chapter. Change to intermediate node. To change a conclusive node into an intermediate node. You will loose the node s conclusion text. This action is irreversible. Therefore, you are prompted for confirmation. Change to conclusive node. To change an intermediate node into a conclusive node. You will loose some of the node s settings, such as the SQL Query, and the branches that are leaving this node. This action is irreversible. Therefore, you are prompted for confirmation. View menu The View menu contains the following commands: Toolbar. To enable/disable the toolbar at the top of the window. Status bar. To display/hide the status bar at the bottom of the window. The status bar displays name and location of the current model, date and time. Notes and Warnings. To enable/disable the Notes and Warnings pane at the bottom of the window. Warnings Only. To display only warnings in the Notes and Warnings pane; notes are hidden. Warnings block the execution of the model. Notes are just for information. Object Details. To enable/disable the Object Details pane at the right side of the window. Straighten all Branches. To straighten segmented branches. Refresh. Refreshes the content in the object detail pane. 2-11

28 To create diagnostic models Graphical Options. Starts the Graphical options dialog. Here, you can change the appearance (colors, shapes, fonts, and so on) of nodes and branches, and the background color of the drawing pane. You can define your own settings and save them as default. The new default settings will be used for all new models. The existing models retain their own settings. The following figure shows the Graphical options dialog. Figure 12 Graphical options dialog 2-12

29 To create diagnostic models Insert menu The Insert menu contains the following commands: intermediate node conclusive node Branch to Connect Nodes Use these commands to add an intermediate node, a conclusive node, or a branch to your model. For details, refer to the command descriptions in the The toolbar section in this chapter. Window menu The upper part of the Window menu contains the following commands: Cascade Tile Horizontal Tile Vertical Use these commands to arrange model subwindows if you view/edit multiple models simultaneously. The lower part of the Window menu shows the model(s) that you have opened. You can use this to switch from one model to another Help menu The Help menu contains the following command: About. Shows version information for the Modeler. 2-13

30 To create diagnostic models The context menu If you right-click in the drawing pane, the context menu is displayed. See the following figure. Figure 13 The context menu The context menu contains various commands to edit the model, such as Start node. Changes the selected intermediate node into a start node, that is: a node where the analysis can start. A model can have multiple start nodes. In the model s default report settings, you can select which of the start nodes is used to start the analysis when you run the model. Straighten selected branch. To straighten a segmented branch. NOTE Copy model diagram to clipboard. To copy the model diagram from the drawing pane to the Microsoft Windows clipboard. This is useful if you want to include the picture in another application, for example you can paste the model chart in an MS Word document. This command only copies the model chart picture, not the underlying settings and syntax. 2-14

31 To create diagnostic models Show Tables used in Queries. Shows the names of the tables that are used in the SQL queries in the model s intermediate nodes. The other commands in the context menu have already been covered in the previous sections in this chapter. For details, refer to The toolbar and The pull down menus in this chapter. 2-15

32 To create diagnostic models 2-16

33 3 General Model Properties For each diagnostic model, you must define a number of basic properties in the Edit Model dialog. Examples of these model properties include: The model s name and descriptive information. Model variables that are used in the model s nodes. The Edit Model dialog The Edit Model dialog is started automatically, when you add a new model. If you want to modify the basic properties of an existing model, for example if you want to add a model variable, you can start the Edit Model dialog in one of the following ways: Select Edit Model Description, Variables, and Connection from the toolbar. Select Model from the Edit menu. Press CTRL and simultaneously press M. 3-1

34 General Model Properties The following figure shows the Edit Model dialog with the basic settings for a diagnostic model. Figure 14 Edit Model dialog with sample model properties The dialog contains the following fields: Name. The name of the model. The name entered here is displayed in the Modeler s title bar. Description. Here, you must enter descriptive information (meta data) about the model. For example: Explain what the model is used for: briefly describe each step (node, branch) in the model chart. Variables. Here you must specify the model variables that will be used in the intermediate nodes and conclusive nodes of your model. Use the + or X buttons at the right side of the variable list, to add or delete variables. Use the Arrow-Up and Arrow-Down buttons to change the order of variables in the list. 3-2

35 General Model Properties NOTE For each variable, you must specify the following parameters: Name. The name of the variable as it is used in the nodes and/or branches of your model. Description. A brief description/explanation. Data Type. You can select the variable s data type from a list. Select Numeric for al numeric variables. Select String for all alphanumeric variables. Storage type (SQL). When running a diagnostic model, you can write the conclusions to a table in an SQL Server database. In this column you can select, per model variable, the SQL Server storage type for the corresponding table field in the conclusion table. Examples of SQL Server storage types include char, varchar, int, decimal and float. If you select do not store from the list, the corresponding variable will not be used as a table field in the conclusion table. For more information, refer to Conclusion table in the To run a diagnostic model chapter. Length. The number of positions available to store the value of the variable. Initial Value Query. Here, you can define an SQL query that calculates relevant initial values for the variable. For example, an SQL query reads the OrderDate field of the transactions in the Sales Order History table to calculate the possible values for the AnalysisYear variable. To create a query, select the Initial Value Query field and subsequently select the button. This will start the Query Builder. For details about the usage of the Query Builder, refer to chapter 7, To define queries. Connection. Here, you must select the desired SQL Server connection from the list. A SQL Server connection is used to link your diagnostic model to the SQL Server database, whose tables you want to analyze. You can add a new SQL Server connection through the SQL Server connections button. Refer to SQL Server Connections in this chapter for details. 3-3

36 General Model Properties SQL Server Connections To define a new SQL Server connection, take one of the following steps: Select SQL Server connections in the Edit Model dialog. Or: Select SQL Server Connections from the File menu in the Diagnostic Modeler s main window. Both actions will start the SQL Server connections dialog that shows a list of existing connections. Figure 15 SQL Server connections dialog In this dialog, select Add to start the Specify SQL Server Connection dialog. See the following figure: Figure 16 Specify SQL Server Connection dialog with sample connection 3-4

37 General Model Properties For each connection, you must specify the server name, login settings, and the database name. Usage of Model variables You can use the model variables, defined in the Edit Model dialog, as context variables (variables that determine the context of an analysis or a conclusion) in all components of the diagnostic model: In intermediate nodes. The context variables are used: To store values that will be transferred, through a branch, to a subsequent node. As parameters in the intermediate node s SQL query; for example, in an expression/calculation in the SELECT clause, or in a condition in the WHERE clause. In branches. Branches are used to transfer values from an intermediate node to a subsequent intermediate or conclusive node: variables in the origin node are mapped to context variables in the destination node. In conclusive nodes. The context variables are used as parameters in the conclusion text. Refer to the following chapters for details about the usage of variables in the components of a diagnostic model: Chapter 4, To define intermediate nodes. Chapter 5, To define conclusive nodes. Chapter 6, To define branches. Chapter 7, To define queries. Chapter 10, Tutorial. 3-5

38 General Model Properties 3-6

39 4 To define intermediate nodes Intermediate nodes are used to specify the analysis that must be done to solve one step in the search process: each intermediate node corresponds with one step in the search process. The analysis in an intermediate node is performed through an SQL Query. EXAMPLE A diagnostic model tries to find an explanation for the below-average turnover of employees in a particular calendar year. The Calculate Avg Values intermediate node contains an SQL query that calculates the average freight costs and discounts for all employees. Another Check Discount intermediate node contains a query that compares the discounts for each bad performing employee to the average discount. If the discounts given by an employee are higher than average, the conclusion Employee <EmployeeName> offered too high discounts to his / her customers. Year of analysis:<analysisyear>. is drawn and the analysis stops. If the discounts are below average, the analysis continues in the next intermediate node. The next Check Freight Costs intermediate node, contains a query that compares the freight costs for each bad performing employee to the average freight cost. If the freight costs for an employee are higher than average, the conclusion The company paid too much freight costs for the sales orders of employee <EmployeeName>. Year of analysis:<analysisyear> is drawn. The following figure illustrates this example. Figure 17 Intermediate nodes, conclusive nodes and branches 4-1

40 To define intermediate nodes NOTE The nodes on the left are intermediate nodes. The nodes on the right are conclusive nodes. You can change the appearance (colors, shapes, fonts, and so on) of nodes and branches in the Graphical options dialog. To start this dialog, select Graphical Options from the View menu or from the context menu. The analysis in an intermediate node is mostly done in a specific context, for example for a particular item or for a particular customer. In each intermediate node, you can define a set of context variables, which determines the context for the node. Context variables in intermediate nodes are often used as parameters in the underlying SQL queries. For example: a query uses a <[year]> parameter. The value for this parameter can be the result of an SQL query in a previous intermediate node or can be user input. Procedure overview To define an intermediate node, take the following steps: 1 Add a new intermediate node in the drawing pane, through one of the following actions: select Add intermediate node from the toolbar, or select Add intermediate node from the context menu, or select intermediate node from the Insert menu, or press <Ctrl> I 2 Drag the node to the desired position in the drawing pane. 3 Enter the intermediate node s properties in the Edit intermediate node dialog. To start this dialog: Double-click the new intermediate node. Or: Select the new intermediate node, and subsequently select Edit selected object from the context menu. Or: Select the new intermediate node, and subsequently select Edit selected object from the Edit menu. Or: Select the new intermediate node and subsequently select Edit in the Object Details pane. Or: select the new intermediate node and press F2. 4-2

41 To define intermediate nodes To define the intermediate node s properties The following figure shows the Edit intermediate node dialog: Figure 18 Edit intermediate node dialog with sample node properties 4-3

42 To define intermediate nodes The dialog contains the following fields: Id. A sequence number that indicates the order in which the nodes have been created: one for the first node, two for the second node, and so on. You can change the id s later on, for example after completing the model. Short title. The text entered here is displayed in the node in the drawing pane. Can be start node. Select this check box if you want to use the node as a start point for your analysis. A diagnostic model can have multiple start nodes. You can select the start node to be used for running the model, and enter initial values for its context variables in the Default Report Settings dialog and in the Diagnostic Analyzer Reporter. A start node should have a different shape compared to a regular intermediate node, for example an octagon instead of a circle. You can change the shape of the start node(s) in the Graphical options dialog. To start this dialog, select Graphical Options from the View menu or from the context menu. Title. The title of the node. The text entered here is displayed as a tool tip, when you position your mouse pointer on the node in the drawing pane. Description. Here you must enter descriptive information (meta data) about the node. For example: explain what the node is used for, briefly describe the values selected/calculated by the query, and so on. Required context. Here you must specify the model variables that form the context of the intermediate node, that is the context variables. You can only select existing model variables from the list. You can add new model variables in the Edit Model dialog. In an intermediate node, context variables are used: As parameters in the intermediate node s SQL query: for example, in an expression/calculation in the SELECT clause, or in a condition in the WHERE clause. And/or: To store values that will be transferred, through a branch, to a subsequent node. Use the + or X buttons at the right side of the variable list, to add or delete variables. Use the Arrow-Up and Arrow-Down buttons to change the order of variables in the list. SQL Query. This area contains four buttons: View. To view the syntax of an existing SQL query. Edit. Starts the Query Builder, where you can design your queries in a graphical way. 4-4

43 To define intermediate nodes NOTE EXAMPLE EXAMPLE EXAMPLE Copy. To copy a query to the Modeler s internal clipboard. From this clipboard, you can paste the query to another intermediate node (in the same diagnostic model, or in a different model). Paste. To copy a query from the Modeler s internal clipboard into an intermediate node. Refer to Usage of SQL Queries for more information. Priority value. Running a diagnostic model renders a weighted list of conclusions: a weight is calculated for each conclusion. The total of all weights is 1. Use this area to define how the weights of the conclusions will be calculated: Same for all records. Each conclusion has the same weight. The total of all weights is 1. So, the weight for each conclusion is 1/<no of conclusions>. For example: if the conclusion sheet contains 100 conclusions, each conclusion has a weight of Based on Priority field. The weight of each conclusion is calculated based on the value of the selected priority field. A priority field is a field/column, in the result set of the intermediate node s SQL query, which quantifies the problem you want to analyze in your diagnostic model. The ValueBelowTarget, in a diagnostic model that analyzes why some employees have a sales turnover that is below average, or the NoOfDaysLate, in a model that analyzes why some deliveries have a lateness that is above average. The calculated weight indicates the relative contribution of each conclusion to the problem that is analyzed. The total of all weights is 1. The intermediate node s SQL Query calculates the ValueBelowTarget for each employee. The ValueBelowTarget is used as priority field, to calculate the weights for the conclusions in the result sheet. Suppose, there are three employees A, B, and C with a sales turnover that is below average. The ValueBelowTarget for these employees is 20,000, 16,000, and 14,000. The calculated weights are 0.40, 0.32, and 0.28 respectively. The total weight is 1. Threshold. The value entered here functions as a filter for further analysis: only records where the priority variable is greater than the threshold will be further analyzed in a subsequent node. The default threshold value is 0. The intermediate node s SQL Query calculates the ValueBelowTarget for each employee. Suppose the query result (Employee / ValueBelow Target) is: A/20000, B/16000, C/14000, D/0, E/-4000 (4000 above target). If the threshold is 0, only employee A, B, and C will be analyzed in the next node. 4-5

44 To define intermediate nodes However, if the threshold is 15,000, A and B are the only records that will be further analyzed. Branch field. You can only select a branch field, if at least two branches start in the current intermediate node. A branch field is a field/column, in the result set of the SQL query, which is used in the branches that start in the current intermediate node. In each branch you can define a branch condition that determines when the branch will be followed. EXAMPLE The Check Discount intermediate node calculates the discount ratio per sales representative. The branch field in this node is DiscountRatio. The node is connected to: A Discount Too High conclusive node. The corresponding branch is followed if DiscountRatio is greater than 1. The conclusion Employee gives too high discounts is drawn. NOTE An Check Freight Costs intermediate node. The corresponding branch is followed if DiscountRatio is less than or equal to 1. The analysis is continued in the Check Freight Costs intermediate node. You can select the desired branch field and the corresponding data type (float, integer, or string) from the pull down lists. For details about branches and branch conditions, chapter 6 To define Branches. 4-6

45 To define intermediate nodes Usage of SQL Queries To define SQL queries To create an SQL query in an intermediate node, select Edit in the Edit intermediate node dialog. This will start the Query Builder (Build SQL Query dialog). See the following figure for an example. Figure 19 Query Builder with sample SQL query Use the Query Builder to create SQL Queries in a graphical way. You can: Select tables from a list. Select fields from the table definition(s) displayed in the upper pane. Join table fields, using the mouse. Define expressions and conditions. And so on. The generated SQL syntax is displayed in the lowest pane of the Query Builder window. This pane is read-only. 4-7

46 To define intermediate nodes NOTES For details about the Query Builder, refer to chapter 7, To define queries. For details about Microsoft SQL syntax, refer to SQL Server Books Online. To view SQL Query syntax To view the syntax of an SQL query in an intermediate node, select View in the Edit intermediate node dialog. This will start the SQL Query window. The View command only shows the SQL query syntax; it does not connect to the SQL Server database. See the following figure for an example. Figure 20 SQL Query window with sample query The syntax displayed is read-only. You can use the Copy button to copy the SQL Query syntax to the Microsoft Windows clipboard. This is useful if you want to include the syntax in another application, for example, you can paste the SQL query syntax in an MS Word document. 4-8

47 5 To define conclusive nodes A conclusive node represents a conclusion that can be drawn based on the analysis in an intermediate node. A conclusion is often specific for a combination of context variables, for example Supplier and Year: Supplier X had a bad delivery performance in year Y. EXAMPLE The Check Discount intermediate node contains a query that compares the discounts for each employee to the average discount. This node is connected to: - a Too High Discounts conclusive node, and - a Check Freight Costs intermediate node. If the discounts given by an employee are higher than average, the analysis process follows the branch to the conclusive node: the conclusion Employee <EmployeeName> offered too high discounts to his / her customers. Year of analysis:<analysisyear>. is drawn and this path of the analysis process stops. If the discounts are below average, the branch to the Check Freight Costs intermediate node is followed and the analysis process continues. The following figure illustrates this example. Figure 21 Intermediate nodes, branches and conclusive nodes NOTE The nodes on the left are intermediate nodes. The nodes on the right are conclusive nodes. You can change the appearance (colors, shapes, fonts, and so on) of nodes and branches in the Graphical options dialog. To start this dialog, select Graphical Options from the View menu or from the Context menu. 5-1

48 To define conclusive nodes The context of a conclusive node, that is the node s context variables, is used to turn a generic potential conclusion into a concrete problem cause. For example, one conclusive node can model the generic potential conclusion Supplier <SupplierId> delivers item <ItemId> unreliably in month <MonthId>. This conclusive node has a required context with three model variables {supplier, item, month}. When the search process visits this conclusive node, it will make the potential conclusion concrete by specifying values for supplier, item, and month and assigning a weight to this conclusion. NOTE Procedure overview To define conclusive node, take the following steps: 1 Add a new conclusive node in the drawing pane, through one of the following actions: select Add conclusive node from the toolbar. Or: select Add conclusive node from the context menu. Or: select conclusive node from the Insert menu. Or: press <Ctrl> J 2 Drag the node to the desired position in the drawing pane. 3 Enter the conclusive node s properties in the Edit conclusive node dialog. To start this dialog: Double-click the new conclusive node. Or: Select the new conclusive node, and subsequently select Edit selected object from the context menu. Or: Select the new conclusive node, and subsequently select Edit selected object from the Edit menu. Or: Select the new conclusive node, and subsequently select Edit in the Object Details pane. Or: Select the new conclusive node and press F2. After creating a conclusive node, you must define at least one incoming branch: the conclusive node must be connected to at least one intermediate node. 5-2

49 To define conclusive nodes To define the conclusive node s properties The following figure shows the Edit conclusive node dialog: Figure 22 Edit conclusive node dialog with sample conclusion The dialog contains the following fields: Id. A sequence number that indicates the order in which the nodes have been created: 1 for the first node, 2 for the second node, and so on. You can change the id s later on, for example after completing the model. Short title. The text entered here is displayed in the node in the drawing pane. 5-3

50 To define conclusive nodes NOTE EXAMPLE Node title. The title of the node. The text entered here is displayed as a tool tip, when you position your mouse pointer on the node in the drawing pane. Start conclusion text with. The first part, before the first variable, of the conclusion text that will be displayed in the conclusion sheet. Context variables. Use this area to design the second (last) part of the conclusion text. You must specify the variables and the text that will be used in the conclusion text. The resulting conclusion text is displayed in the Conclusion text field. You can only select existing model variables from the list. You can add new model variables in the Edit Model dialog. Use the + or X buttons at the right side of the variable list, to add or delete variables. Use the Arrow Up and Arrow Down buttons to change the order of variables in the list. This also changes the resulting conclusion text. Nr. of significant digits floating point variables. Here you must specify the number of digits to which all floating point variables in the conclusion text will be rounded. For example: the conclusion text contains the average discount percentage variable. The value of this variable, calculated through an SQL query in an intermediate node, is If the Nr. of significant digits floating point variables is 2, the conclusion text will display a percentage of Conclusion text. This field is read-only. It displays the conclusion text as a whole. See the previous figure for an example. During the execution of a diagnostic model, the search process makes the potential conclusion concrete: when visiting a conclusive node, the search process specifies values for the context variables, for example <EmployeeName>, <EmployeeExperience>, and <AnalysisYear>, and assigns a weight to the conclusion. At the end of the search process, the total weight of all conclusions will be 1. The following table contains two examples of concrete conclusions: Sample conclusions Node Id Conclusion Weight Little Experience Employee Ms. Anne Dodsworth (Sales Representative) has not much working experience (21 months). Year of analysis: Too High Discounts Employee Mr. Steven Buchanan (Sales Manager) offered too high discounts to his / her customers. Year of analysis:

51 6 To define branches Branches (arrows) are used to connect nodes: they determine the direction in the search process. Branches transfer values from an intermediate node to a subsequent intermediate or conclusive node: variables in the origin node are mapped to context variables in the destination node. You can define multiple branches to link a single intermediate node to various subsequent intermediate nodes and/or conclusive nodes. If multiple branches start in the same intermediate node, you can define a condition for each branch. A branch condition determines when the branch will be followed. See the following example: EXAMPLE The Check Discount intermediate node calculates the discount ratio per sales representative. This node is connected to: A Too High Discounts conclusive node. The corresponding branch is followed if the discount ratio is greater than 1. The conclusion Employee gives too high discounts is drawn. An Check Freight Costs intermediate node that calculates the freight cost ratio per sales representative. The corresponding branch is followed if the discount ratio is less than or equal to 1. The following figure illustrates this example. Figure 23 Intermediate nodes, branches and conclusive nodes The nodes on the left are intermediate nodes. The nodes on the right are conclusive nodes. 6-1

52 To define branches NOTE You can change the appearance (colors, shapes, fonts, and so on) of nodes and branches in the Graphical options dialog. To start this dialog, select Graphical Options from the View menu or from the context menu. Procedure overview Take the following steps to add a branch between two nodes: 1 Select the origin node. This is the intermediate node where the branch will start. 2 Select the destination node. This is the intermediate or conclusive node to which the origin node will be connected. The branch to be added is now indicated by a dashed arc. 3 Add a branch between these two nodes: To add a branch: Select Connect Nodes by inserting a Branch from the toolbar. Or: Select Branch to Connect Nodes from the Insert menu. Or: Select Connect Nodes with a Branch from the context menu. Or: Type <Ctrl> b. 4 Define the branch s properties in the Edit Branch dialog. To start this dialog Double-click the new branch. Or: Select the new branch, and subsequently select Edit selected object from the context menu. Or: Select the new branch, and subsequently select Edit selected object from the Edit menu. Or: Select the new branch, and subsequently select Edit in the Object Details pane. Or: Select the new branch and press F2. 6-2

53 To define branches To define the branch s properties The following figure shows the Edit Branch dialog: Figure 24 Edit Branch dialog with sample branch properties 6-3

54 To define branches The dialog contains the following fields: Id. A sequence number that indicates the order in which the branches have been created: one for the first branch, two for the second branch, and so on. You can change the id s later on, for example after completing the model. Short title. The text entered here is displayed in the drawing pane. NOTE EXAMPLE Description. Use this field to enter meta data, for example a description of the function/purpose of the branch. The text entered here is shown in the HTML page that you can display through the Show model an html command. Branch Condition. Use this area to specify the condition under which the search process will follow this branch. You can only enter a branch condition if at least two branches start in the same intermediate node. A branch condition is always based on the branch variable that is specified in the origin node (the intermediate node where the branch starts). You can enter two types of branch conditions: Single Value: you can specify only one value. The analysis process will follow this branch only for records where the branch variable equals the value specified here. For the other records another branch will be followed. For example: the branch variable is OrderValue. The single value specified in the branch condition is The analysis process will follow this branch only for records where the OrderValue = Value Range: the analysis process will follow this branch only for records where the branch variable matches the specified range. You can define the range s lower and upper boundary and the corresponding operators, < or <=. The resulting condition is displayed in the Condition field. For example: the branch variable is DiscountRatio. The following condition is specified: 1 < DiscountRatio < No upper boundary. The analysis process will follow this branch only for records with DiscountRatio greater than 1. Branches that start in the same origin node must always exclude each other. Therefore, the branch conditions must be complementary. Two branches start in the Check Discount intermediate node. The first branch, Excessive Discount, is followed for records with DiscountRatio greater than 1. The second branch, Acceptable Discount is followed for all other records, that is records with DiscountRatio less than or equal to

55 To define branches Context Mapping. Use this area to link origin variables to destination variables. Origin variables are variables that belong to the branch s origin node (the node where the branch starts). Destination variables are context variables in the branch s destination node. At runtime, the analysis process will use the context mapping to transfer the values of origin variables to the corresponding context variables in the destination node. You can map two types of origin variables: Origin (Context). These are the context variables of the origin node. For example: the FreightCostRatio variable that is calculated in one of the previous intermediate nodes, and subsequently transferred to the origin node. Origin (Results). These variables contain the results of the origin node s SQL query. For example: the DiscountRatio variable that is calculated in the SQL query s SELECT clause. To map variables, drag from an origin variable to the corresponding destination variable. To delete a link between two variables, right-click the link and select Remove. The following figure shows a sample branch in the Modeler s drawing pane. Figure 25 Excessive discounts branch in model chart 6-5

56 To define branches The Excessive Discounts branch connects the Average Discount intermediate node and the Too High Discounts conclusive node. The following figure shows the corresponding context mapping in the Edit Branch dialog. Figure 26 Context mapping The origin variables, belonging to the Average Discount intermediate node are displayed in the left part of the Context Mapping area. The destination variables, belonging to the Too High Discounts conclusive node are displayed in the right part of the Context Mapping area. The origin variables EmployeeName and AnalysisYear are mapped to context variables in the destination node. These context variables are used as parameters in the destination node s conclusion text. 6-6

57 7 To define queries This chapter covers a number of features of the Query Builder (Build SQL Query dialog). The Query Builder is a functionality that enables you to create SQL queries in a user-friendly, graphical way. To start the Query Builder In the Diagnostic Analyzer Modeler, you can use the Query Builder to define the SQL query for an intermediate node. To start the Query Builder, complete the following steps: 1 Open an intermediate node in the Modeler s drawing pane. This will start the Edit intermediate node dialog. 2 Select Edit in the Edit intermediate node dialog. 7-1

58 To define queries Now, the Query Builder main window is displayed, as shown in the following figure. Figure 27 Query Builder main window with sample query To define a query, complete the following steps: 1 Select the desired table(s) from the table list in the upper pane. The tables are included in the FROM statement of the query. 2 Define joins (optional). 3 Select the desired table fields. These fields are included in the SELECT statement of the query. 4 Define expressions (optional). These expressions are included in the SELECT statement of the query. 5 Define conditions (optional). These conditions are included in the WHERE statement of the query. The steps mentioned here are described in more detail in the following sections. 7-2

59 To define queries NOTE The syntax of the generated SQL Query is displayed in the Preview SQL Query pane. You can preview the query results by clicking Fire the Query (the exclamation mark button). See the To preview the query results section in this chapter for details. To check the syntax of the query, select Check Syntax from the tool bar in the Preview SQL Query pane. The Query Builder supports the usage of sub queries. Refer to To define sub queries in this chapter for more details. To select tables You can select the desired tables and/or views from the tables list in the upper left pane of the Query Builder main window. To select a table or view, take one of the following steps: Select the desired table/view and click >. Double-click the desired table/view. Now, the table or view is displayed in the upper pane of the Query Builder window, as shown in the following figure. Figure 28 Table list with selected table NOTE The S button in the table-list dialog box is used to define sub queries. If you select this button, a separate Sub query window is started. For more details, see To define a subquery. 7-3

60 To define queries To define joins (optional) To join two tables, complete the following steps: 1 Select a field to be joined in the first table. 2 Drag the mouse pointer to the related field in the other table. 3 Repeat this procedure for each field that must be included in the join. 4 Right-click a join to define the join type. A menu with the following options is displayed: Remove. Use this command to remove a join. Select All From <the left table>. Use this option to define a left outer join. NOTE Select All From <the right table>. Use this option to define a right outer join. If you choose Select All From <the left table> and Select All From <the right table> from the menu, a full outer join is generated. This join type selects all records from both tables. If you do not select any option from the menu, an inner join is generated. This join type selects only those records from both tables, for which the joined fields contain the same values. Records that do not have a matching record in the other table are ignored. For details about the different join types, refer to the SQL Server Books Online. See the following figure for an example. Figure 29 Upper pane of Query Builder window with two joined tables NOTE For slowly-changing dimension tables in an ibaan Decision Manager data warehouse database, you must define joins on surrogate key fields. The fact tables in the data warehouse contain various surrogate key fields that can be joined to the surrogate key fields of the dimension tables 7-4

61 To define queries See the ETL Modeler online Help for details about slowly-changing dimensions. To select table fields To select a table field, simply double-click the field in the table in the upper pane of the Query Builder window. As a result, the field is displayed in the Selected Fields pane. See the following figure. Figure 30 Selected Fields pane Some features of this pane are: To remove a selected field, click Delete (the last button in the button bar). NOTE To change the order of the selected fields, use the arrow up/down buttons on the button bar. If you click the Aggregation column of a selected field, you can select aggregation functions and other statistical functions from a list. The list contains various functions, such as AVG, COUNT, STDEV, SUM and VAR. For details about the usage of the functions you can select, refer to the SQL Server Books Online. To define expressions (optional) Use this feature if you want to include a calculation or formula in the select statement of the query. To define a new expression, select Add Expression (+) from the toolbar in the Expressions pane. To edit an existing expression, select Edit Expression (second button on the toolbar). 7-5

62 To define queries Both actions will start the Build Expression dialog box. See the following figure. Figure 31 Build Expression dialog box and Expressions pane NOTE To define an expression, complete the following steps: 1 Select the desired table field, and, optionally, an aggregation function or statistical function in the Existing Fields pane. You can select various functions, such as AVG, COUNT, STDEV, or SUM. 2 Optionally: select other SQL functions in the T-SQL Functions pane. You can select various Date and Time, Mathematical and String functions, for example: DATEADD, GETDATE, ROUND, SQRT, LTRIM and UPPER. You can use the pull down list in the Parameters field to include context variables in the expression. 3 Define a variable name, to which the expression will be assigned. 4 Click Insert to insert the resulting expression in the lowest pane of the Build Expression dialog box. If necessary, edit the expression in the lowest pane. 5 Select OK to include the expression in your SQL Query. The expression is displayed in the query syntax in the lowest pane of the Query Builder main window. For details about the usage of the functions you can select, refer to the SQL Server Books Online. 7-6

63 To define queries To define conditions (optional) You can use conditions to filter the results of the query; for example, only select customers with country code = USA. In the lowest pane of the Query Builder window, conditions are included in the WHERE clause of the query syntax. To define a new condition, select Add Condition (+) from the toolbar in the Conditions pane. To edit an existing condition, select Edit (second button on the toolbar). Both actions will start the Build Condition dialog box See the following figure. Figure 32 Build Condition dialog box with Conditions pane To define a condition, complete the following steps: 1 Select the desired table field from the list. For example, select the OrderDate field in the Orders table (Orders.OrderDate). 2 Select the desired comparison operator. You can select the following operators: =, <, >, <=, >=, <>, LIKE and NOT LIKE. 3 Enter the value or expression, or select the parameter to which the selected field must be compared. To define an expression, click... This will start the Build Expression dialog box where you can enter the desired expression. The parameters that you can select from the pull down list are the context variables of the current intermediate node. The names of the context variables are enclosed by <[ and ]>, for example <[AnalysisYear]>. 7-7

64 To define queries EXAMPLE 4 Click Insert to insert the resulting expression in the lowest pane of the Build Condition dialog box. 5 Click OK to include the condition in your SQL Query. The expression is displayed in the WHERE clause in the query syntax in the lowest pane of the Query Builder window. 6 Optional: manually enter additional syntax. For example, use the DATEPART function to filter the year from the OrderDate field. See the previous screenshot Some other features of the Build Condition dialog box are: You can disable/enable a condition through the Condition is active check box. The C button clears your condition. The Append with OR button enables you to combine multiple comparisons with an OR statement. This example is based on the Customers table in the Northwind database. You want to select customer records, where the Country code is USA or CAN. First define the first comparison, Customers.Country = USA, and click Insert. Subsequently, define the second comparison, Customers.Country = CAN, and click Append with OR. The resulting expression is Customers.Country = USA OR Customers.Country = CAN. Click OK to include the condition in your SQL Query. The expression is displayed in the WHERE clause in the Preview SQL Query pane. To define subqueries Defining a subquery is useful in various situations where a single query cannot provide the desired information. For example: You want to select all items that have a sales price that is higher than average. A subquery selects the average sales price. Subsequently, the main query selects all items with a price that is higher than this average price. You want to calculate the average order value for all employees. A subquery calculates the total order value for each employee. Subsequently, the main query calculates the average of these totals. 7-8

65 To define queries To define a subquery, select Define Sub Query (S) in the table list in the upper pane of the Query Builder window. This will start the Sub Query dialog box as shown in the following figure. Figure 33 Sub Query dialog activated from Query Builder main window NOTE EXAMPLE The Sub Query dialog box is similar to the Query Builder main window. For a description of the panes in this dialog box, refer to the descriptions of the corresponding Query Builder panes in this chapter. In the Name field, you must enter the name of the virtual table to which the subquery's results will be assigned. After saving the subquery, the name of this virtual table is displayed in the table list in the Query Builder main window. See the following figure for a sample subquery. 7-9

66 To define queries Figure 34 Sample subquery The subquery in the screenshot calculates the total order value (variable: TotalOrderValue) for each employee. These totals are stored in a virtual table OrderValueByEmployee. Subsequently the main query calculates the average of these totals. See the following figure. 7-10

67 To define queries Figure 35 Main query calculates the average order value per employee The subquery s virtual result table is displayed at the top of the table list. The resulting SQL syntax of the main query, including the subquery, is: SELECT AVG(OrderValueByEmployee.TotalOrderValue) AS AverageOrderValue FROM ( SELECT Orders.EmployeeID AS EmployeeID, SUM(([Order Details].Quantity*[Order Details].UnitPrice*(1-[Order Details].Discount))-Orders.Freight) AS TotalOrderValue FROM dbo.orders AS Orders INNER JOIN dbo.[order Details] AS [Order Details] ON (Orders.OrderID = [Order Details].OrderID) WHERE (DATEPART(yyyy,Orders.OrderDate) = <[AnalysisYear]>) GROUP BY Orders.EmployeeID ) AS OrderValueByEmployee 7-11

68 To define queries NOTE To edit an existing subquery, first remove the subquery s virtual result table from the upper pane in the main query. Subsequently, click S to open the subquery. To preview the query results You can preview the query results by clicking Fire the Query (the exclamation mark button) in the toolbar of the Preview SQL Query pane. If model variables are used in the query, the Parameter Values dialog is started. See the following figure. If no variables are used, the query results are displayed directly. Figure 36 Parameter Values dialog 7-12

69 To define queries In this dialog, you must enter the values for the context variables that are used as parameters in the query. Click OK to view the query results. See the following figure. Figure 37 Sample query results The query results are displayed in the Preview Query Results window. This window contains the following fields and buttons: Nr. rows shown. To specify/change the maximum number of rows to be displayed. If you change this setting, select Refresh to get the desired number of rows. Refresh. To refresh the query results. Use this if you changed the value in the Nr. rows shown field. Copy. Copies the query results to the Microsoft Windows clipboard. This is useful if you want to include the query results in another application, for example in an MS Excel spreadsheet. Close. To close the Preview Query Results window. 7-13

70 To define queries 7-14

71 8 To run a diagnostic model You can run a diagnostic model through: the Execute the Model command in the Diagnostic Analyzer Modeler. Before using this command, you must enter a number of default report settings. The Diagnostic Analyzer Reporter. You can start the Reporter from the ibaan Diagnostic Analyzer menu, and through the ibaan Decision Manager Extraction Job Runner. The latter enables you to schedule an automated analysis, for example directly after each data warehouse upload. The search results are displayed in a conclusion sheet. This is a weighted list of to-the-point conclusions that can directly aid the decision making process. You can also save intermediate search results and analyze these through the Tree Analyzer. Additionally, you can save the conclusions in an SQL Server table, and define an MSOLAP cube through which you can analyze the search results in a flexible way. To define the default report settings for running a model Before you can run a diagnostic model through the Execute the Model command in the Modeler, you must enter a number of default report settings in the Default Report Settings dialog. To start this dialog: select Edit default report settings from the tool bar. Or: select Default report settings from the Edit menu. 8-1

72 To run a diagnostic model The following figure shows the Default Report Settings dialog: Figure 38 Default Report Settings dialog The dialog contains the following fields: Diagnostic Analysis Model. This read-only field displays the full path and the name of the current model. Conclusion sheet. The full path and the name of the conclusion sheet that is generated when you run the model. The conclusion sheet is stored as an.htm file. You can use the button to select a folder through the Location Index Conclusion Sheet dialog box. If you select Show Report after model run, the conclusion sheet is displayed automatically after running the model. You can select Pictures to include a picture, for example your company logo, in the upper left corner and/or upper right corner of the conclusion sheet. Start Node. A model can contain multiple potential start nodes. Select the desired start node from the list and enter the initial values for the corresponding context variables. 8-2

73 To run a diagnostic model NOTE EXAMPLE In the Diagnostic Analyzer Modeler, you can change any existing intermediate node to a potential start node: open the desired intermediate node and subsequently select Can be start node in the Edit intermediate node dialog. Parameters and Settings. You can enter various parameters that have an impact on the output that will be generated: Tolerance. The maximum cumulative weight of the conclusions that will be displayed in the conclusion sheet. The analysis process automatically stops when this value is reached. If you enter 1, all records will be processed and all corresponding conclusions are displayed. Use this parameter to improve the performance and to reduce the number of conclusions displayed. For example: if you enter 0.8, only the conclusions that represent 80% of the analyzed problem are displayed. Maximum number of processed records. Another parameter that you can use to improve the performance of the search process. In addition to the Tolerance explained above, it may also be wise to set a maximum to the number of records in a result set that will be analyzed. Reducing this number will reduce the processing time and the number of conclusions generated. Maximum number of nodes in Result tree. When you run a model, each intermediate node usually generates various successor nodes. A diagnostic model, that contains a lot of intermediate nodes, can generate an excessive number of result nodes. See the following example. Use this parameter to limit the number of result nodes generated by the search process. Consider the following: A diagnostic model contains 4 intermediate nodes: A, B, C, and D. During execution of this model, start node A selects 52 records. 21 of these records are analyzed further in intermediate node B: so, node A generates 21 successor nodes. The first of these successor nodes generates 12 successor nodes. The second one generates 8 successor nodes, and so on. In total, the 21 successor nodes generated by node A, result in 180 successor nodes. These 180 nodes generate 1,552 successors. These, in turn, generate 14,092 successors. And so on. NOTE Situations such as this one can occur; they are not always usual. 8-3

74 To run a diagnostic model NOTE Result Tree. To save intermediate results, select the Save intermediate results (Result Tree) to check box and specify the full path and the name of the intermediate result file (.drr file) to be generated when you run the model. You can use the button to select a folder. You can analyze the intermediate results through the Tree Analyzer. Refer to chapter 9, To analyze the Result tree for details. If you select the Also save the result sets (query results) check box, the query results for each intermediate node will also be saved to the intermediate result file (.drr file). The query results are displayed in the Result Set pane of the Tree Analyzer. You can only save the intermediate results if you did not select the Only save conclusions check box. Conclusion table. You can save the conclusions in an SQL server table. This enables you to define an MSOLAP cube, based on this table, through which you can analyze the search results in a flexible way. To save the conclusions to an SQL Server table, you must: Select the desired SQL Server connection from the list. Enter the name for the Conclusion Table, by default: DA_<Model name>. Each conclusion is a separate record in this table. Enter the name for the Run Id Table, by default: DA_<Model name>_d. This table contains one record for each run id. You can join both tables on the RunId field. 8-4

75 To run a diagnostic model To run a model from the Modeler To run a model, select Execute the Model from the tool bar. This option uses the defaults specified in the Default Report Settings dialog, connects to the SQL Server database to which your model is linked and executes the search process. The conclusion sheet is displayed automatically. See the following figure for a sample conclusion sheet. Figure 39 Sample conclusion sheet To run a model through the Reporter Take the following steps to run a diagnostic model through the reporter: 1 Start the Reporter from the ibaan Diagnostic Analyzer menu. 2 Enter the required parameters in the Diagnostic Analyzer Reporter dialog. 3 Execute the selected model through the Execute Search command. 8-5

76 To run a diagnostic model The Diagnostic Analyzer Reporter dialog The following figure shows the Diagnostic Analyzer Reporter dialog. Figure 40 Diagnostic Analyzer Reporter dialog This dialog is almost identical to the Default Report Settings dialog. Therefore, only the differences compared to that dialog are described. Refer to the description of the Default Report Settings dialog in the previous section, for information about the remaining fields and buttons. Diagnostic Analysis Model. Here, you must enter the full path and the name of the diagnostic model to be executed. If you select Load Default Settings, the default report settings of the selected model are loaded. This will overwrite the report settings you entered so far. SQL Server connections. You can use this button to add a new SQL Server Connection. Subsequently, you can select the new connection from the list, and store the conclusion table in the corresponding SQL Server database. New Report Definition (File). Use this command to create a new report definition. You can start this command from the tool bar and from the File menu. Alternatively, you can type <Ctrl> N. 8-6

77 To run a diagnostic model Open Report Definition (File). Use this command to load the content of a report definition file (.dar file). This will overwrite the report settings you entered so far. You can start this command from the tool bar and from the File menu. Alternatively, you can type <Ctrl> O. Save Report Definition (File). To save the settings you entered in a report definition file (.dar file). You can start this command from the tool bar and from the File menu. Alternatively, you can type <Ctrl> S. Save Report Definition as. To save the current settings in a new report definition file. You can enter the name for the new file in the Save Report Definition dialog box. You can start this command from the File menu. Execute Search. To run the selected diagnostic model with the settings you specified. You can start this command from the tool bar and from the Execute menu. Alternatively, you can type <Ctrl> R. Execute a model through the Extraction Job Runner If you use ibaan Decision Manager, you can automate the execution of your diagnostic models through the Extraction Job Runner. For example, you can schedule the execution of your diagnostic models after each data warehouse upload. Take the following steps to automate the execution of a diagnostic model: 1 Generate a report definition file. 2 Include the report definition file in an extraction job. You can generate a report definition file through the Diagnostic Analyzer Reporter: start the reporter, select the desired model and enter the desired parameters. Subsequently, save the report settings in a report definition file (.dar) file. Subsequently, include the report definition file in an extraction job: 1 Start the Extraction Job Runner from the ibaan Decision Manager/System Management menu. 2 Add a new job, or open an existing one. 3 Select Insert Job Line. This starts the Job Line Details dialog. Select Diagnostic Analyzer in the Jobline type field. Click Options and select the desired report definition file (.dar file). 8-7

78 To run a diagnostic model 4 Save the job: select Save from the File menu in the Extraction Job Runner main window. 5 Generate a batch file that executes the job: select Make Batch File from the File menu in the Extraction Job Runner main window. See the following figures for an example: Figure 41 Job Line Details dialog 8-8

79 To run a diagnostic model Figure 42 Extraction Job Runner with sample job data NOTE To automate the execution of a job, you must include the job s batch file in a Scheduled Task. You can select the batch file through the Browse button in the Scheduled Task Wizard. To start this wizard, select Scheduled Tasks from the Administrative Tools (Common) menu and subsequently select Add Scheduled Task. Output Running a diagnostic model can result in the following output: A conclusion sheet. An intermediate results file. A conclusion table. 8-9

80 To run a diagnostic model Conclusion sheet The conclusion sheet is a weighted list of to-the-point conclusions that can directly aid the decision making process. The conclusion sheet is saved as an.htm file. See the following figure for an example: Figure 43 Sample conclusion sheet Intermediate results file You can also save intermediate search results in a.drr file and analyze these through the Tree Analyzer. Refer to chapter 9, To analyze the Result tree for details. Conclusion table Additionally, you can save the conclusions in an SQL Server table, and define an MSOLAP cube through which you can analyze the search results in a flexible way. The output of the model run is written to two tables. The Conclusion Table. The default name for this table is: DA_<Model name>. Each conclusion is a separate record in this table. This table can function as a fact table on which you can base an MS OLAP cube. Enter the name for the Run Id Table. The default name for this table is: DA_<Model name>_d. This table contains one record for each run id. You can join both tables on the RunId field. 8-10

81 To run a diagnostic model The following figure shows the table definition of a sample conclusion table: Figure 44 Table definition sample conclusion table The columns RunId, NodeId, NodeTitle, ConclusionId and Conclusion exist in every conclusion table. The other columns however, AnalysisYear, EmployeeId, and so on, depend on the structure of the diagnostic model that was run: these are the model variables that are specified in the Edit Model dialog. 8-11

82 To run a diagnostic model See the following figure: Figure 45 Edit Model dialog with variables that are used in the conclusion table For each model variable, you can define the SQL Server storage type and the length for the corresponding table field that will be generated in the conclusion table. If you select -do not store- in the Storage type (SQL) field, the corresponding variable is not used as a table field in the conclusion table. The following figure shows records in a sample conclusion table: Figure 46 Sample conclusion records 8-12

83 9 To analyze the Result tree Running a diagnostic model renders a conclusion sheet: a weighted list of to the point conclusions that can directly aid the decision making process. In addition, you can choose to generate a Result Tree file when running the model. A Result Tree file (drr file) contains all intermediate steps that led to the conclusions. You can analyze the contents of the Result Tree file in full detail through the Tree Analyzer. Report settings To make sure that a Result Tree file will be generated during the execution of a diagnostic model, you must: Select the Save Intermediate Results (Result Tree) to check box. Specify the name and location for the Result Tree file (drr file). This is done in the Modeler s Default Report Settings dialog or in the Diagnostic Analyzer Reporter. See the following figure for an example. These settings are taken into account when you execute the diagnostic model. 9-1

84 To analyze the Result tree Figure 47 Diagnostic Analyzer Reporter with settings for generating an intermediate results file To analyze the Result tree To analyze the contents of an intermediate results file, take the following steps: 1 Start the Tree Analyzer through the ibaan Diagnostic Analyzer menu. 2 Select Open in the Tree Analyzer s Control Panel. This will start the Open Analysis Results File dialog. 3 Select the desired results file (drr file) and click Open. 9-2

85 To analyze the Result tree Now, the contents of the results file are displayed in the Result Tree Analyzer window. See the following figure for an example. Figure 48 Tree Analyzer with Control Panel Tree Analyzer structure The Result Tree Analyzer window consists of the following components: A Tree Diagram pane where the nodes and branches of the Result tree are displayed. Two types of result nodes can be distinguished: intermediate result nodes and conclusive result nodes. The appearance (colors, shapes, fonts, and so on) of nodes and branches in the Tree Diagram pane is determined by the settings that were specified in the Graphical options dialog in the Diagnostic Analyzer Modeler. An object detail pane: this pane consists of two parts: The upper part shows the tree diagram layout. Here, you can see the size and position of the displayed part of the Result tree, compared to the tree diagram and the tree diagram pane as a whole. 9-3

86 To analyze the Result tree NOTE The lower part shows important details about the node that is selected in the Tree Diagram pane. For example, the context variables and the variables and conditions of the SQL query of the selected intermediate node are displayed. A Result Set pane. If you select an intermediate result node in the Tree Diagram pane, the result set of the node s SQL query is displayed here. See The Result Set Pane for details. The results are only displayed if you selected Also save the result sets (query results) in the in the Modeler s Default Report Settings dialog or in the Diagnostic Analyzer Reporter. A Control Panel with various options to determine which information will be displayed. The Control Panel is a separate window that can be moved independently from the Tree Analyzer main window. See The Control Panel for details. The appearance (colors, shapes, fonts, and so on) of nodes and branches in the Tree Diagram pane is determined by the settings that were specified in the Graphical options dialog in the Diagnostic Analyzer Modeler. 9-4

87 To analyze the Result tree The Result Set Pane The Result Set pane displays the query results of the intermediate node that is selected in the Tree Diagram pane. See the following figure for an example. Figure 49 Sample query results The figure shows the query results of the Find Bad Performers node. For each record, the following information is displayed: Id Target Node. The id number of the target result node in which the record will be analyzed further. In the screenshot, the target result nodes are the Employee Experience nodes. In the Tree Diagram pane, the corresponding id numbers are displayed near the branches leading to these target nodes. 9-5

88 To analyze the Result tree Weight. The calculated weight of the record. This weight indicates the relative contribution of each record to the problem that is analyzed. The total of all weights is 1. The weight calculation is usually based on a priority field. If there is no priority field, the weight will be the same for all records. See chapter 4, To define intermediate nodes for details. All Other columns (EmployeeId, ValueBelowTarget, and so on). These are the actual query results. The Control Panel The Control Panel is a separate window that can be moved independently from the Tree Analyzer main window. It contains various options to determine which information will be displayed in the Tree Analyzer window. The following figure shows the Control Panel. Figure 50 The Control Panel 9-6

89 To analyze the Result tree The Control Panel consists of the following areas: Root. Use this area to determine the start point for the Result tree that will be displayed. After changing a setting, you must select Display Tree to refresh the displayed Result tree. First node. This is the default setting when opening a Result Tree file. If you select this, the Result tree will start with the start node that was selected in the Modeler s Default Report Settings dialog or in the Diagnostic Analyzer Reporter. Selected node. Use this to choose another node as the start point for the Result tree to be displayed: select the desired node in the Tree Diagram pane and subsequently select Selected node. Node Selection. Here you can determine the number of nodes to be displayed in the Tree Diagram pane. After changing a setting, you must select Display Tree to refresh the displayed Result tree. Show all successor nodes. For each node, all successor nodes will be displayed. I.e. the complete Result tree is displayed. Selecting this option can result in a huge number of nodes being displayed. Show <number> with highest weight. Use this option to limit the number of nodes displayed in the Tree Diagram pane. For example: you select 5. For each node, only the 5 successor nodes with the highest weights (the top 5 successor nodes) will be displayed. Display Options. This area contains the following options: Show result set of selected node. If you select a node in the Tree Diagram pane, the result set of the corresponding SQL query will be displayed in the Result Set pane. The query results can only be displayed if you selected Also save the result sets (query results) in the in the Modeler s Default Report Settings dialog or in the Diagnostic Analyzer Reporter. Zoom factor. Use this to change the scale of the displayed Result tree. You can select the desired zoom factor from the list. Actions. This area contains the following buttons. Open. To open a Result Tree file (.drr file). Starts the Open Analysis Results File dialog where you can select the desired file. Display Tree. Use this button to refresh the displayed Result tree after changing a layout setting. Show summary. Displays summary information about the selected model, report settings, and results. See the following figure for an example. 9-7

90 To analyze the Result tree Exit. To close the Tree Analyzer. The following figure shows an example of information displayed through the Show summary command. Figure 51 Sample summary information 9-8

91 10 Tutorial In this tutorial, we will describe all the steps that must be undertaken to build a Diagnostic model. The model is also delivered with the software as the Northwind Employee Turnover Analysis model (Northwind.dam). It analyzes data in the Northwind sample database, packed with Microsoft SQL Server. The model The model aims at finding an explanation for the below-average turnover by employees in a particular calendar year. The conclusions that come out of a model run will be employee specific. The employee with the largest gap between target (or average) and their personal turnover (order value) will be on top of the list (will get the highest weight). It must be stressed that the analysis worked out in the model is a good example of the use of Diagnostic models, but is probably not very realistic. It assumes that there is, indeed, an explanation (one of a number of causes) for a turnover that is less than the average or below a certain target. However, it is unlikely that the makers of the Northwind database had such cause-effect relationships in mind. The model has the following features: 1 The analysis and conclusions are specific for one calendar year. 2 In the first two intermediate nodes, we calculate three characteristics for the group of employees as a whole: What was the average percentage discount that was offered by the employees? What was the average amount of freight costs? What was the average turnover per employee? 3 In the third step (node) we compare the performance of each employee with the averages calculated in the first two nodes. Also, the working experience of the employees is calculated. It is possible to skip the first two steps and start directly here. In that case the user must manually enter values for the average discount, the average freight and the average turnover. Such manually entered values are then used as targets and not real averages. 10-1

92 Tutorial 4 During the subsequent steps, the comparison between the group averages and the values for the individual employees is used to arrive at conclusions. Steps to build the model Some overall model properties 1 Start the Modeler application (select Modeler from the ibaan Diagnostic Analyzer menu). 2 Select New from the File menu: an empty model design window will be opened. 3 The Edit Model dialog will pop-up automatically. Specify a database connection, that is select a connection to the Northwind database from the Connection drop-down list. If the drop-down list is empty or does not contain the desired connection, select SQL Server connections. This will start the SQL-Server connections dialog, where you can add (or select) a connection. Select the Test button to test a connection. 4 It is also advised to enter a name for the model and a (short) description. In the Model Variables grid, we must enter all variables that will be used in the context of the model nodes to store and transfer information. At this stage, it is a bit difficult to know exactly which variables we would like to use. The only we know for sure is that the analysis is done for a particular calendar year. Hence, we need a Model variable Analysis Year. See the next section for details. The Model Variable AnalysisYear 1 To add this model variable, select the + button in the Edit Model dialog. 2 Change the name variable1 into AnalysisYear. 3 Add a description in the second column of the grid. 4 AnalysisYear is a numerical variable, so leave the third column as it is. 5 The fourth column (Storage type) is the SQL data-type that will be adopted when the value(s) of the Model Variable will be written back into a SQL server database (together with the conclusions). The smallint satisfies our needs. The fifth column, Length, is not relevant for integer values, and is therefore skipped automatically (this column is, however, relevant for string values). 10-2

93 Tutorial The sixth column indicates whether an initial value query has been defined. An initial value query is an SQL query that may assist in setting the initial value of context variables in start nodes. In the current model, AnalysisYear will be the only context variable in the first intermediate node (one of the two start nodes of the model) and it makes sense to define a query for this variable. This query should return all calendar years for which we have data in our database. 6 Select the ellipsis ( ) button in the Initial value query column, to define the initial value query. This will start the Build SQL Query dialog, with a list of tables (and views) in the Northwind database displayed in the upper left pane of the form. 7 To find out for which calendar years we have sales transactions, we need the Orders table. Double-click on dbo.orders, now this table will appear in the joiner area (the area where tables can be joined and table fields can be selected). 8 The dbo.orders table does not contain a CalendarYear field, and therefore we must create it with an expression: press the + button of the Expressions pane to start the Build Expression dialog. Enter CalendarYear as variable name in the Variable field. 9 In the expression, we need the OrderDate field of the Orders table. This field can be looked-up with the combo-boxes in the Existing Fields frame. Select Orders from the Table list, OrderDate from the Field list and <none> from The Aggregation list, and press Insert. 10-3

94 Tutorial 10 From this field, we must extract the calendar year by using the T-SQL DATEPART function. This can be entered manually in the expression field. The end result must look like: Figure 52 Resulting expression 11 Press OK. 12 Now, we have a valid SQL query. However, it will return the calculated CalendarYear field for all records in the Orders table. To make sure that we get all years only once, select the Unique records only check box on the form. 13 Test the query by pressing the! button on the Preview SQL Query pane. It should return three records with only one field: 1996, 1997, and Click OK to leave the Build SQL Query dialog. 15 Click OK to leave the Edit Model dialog. The first intermediate node In this first node, we will calculate two variables for the group of employees: the average discount and the average freight costs per sales order. Take the following steps to create the first intermediate node: 1 Select intermediate node from the Insert menu, to add an intermediate node to the drawing area. 10-4

95 Tutorial NOTE 2 Double-click on the node to set its properties. The Edit intermediate node dialog will appear. Enter a short-title ( Av. Discount and freight ), a title, and a description. 3 Since the node will act as one of the start nodes of the model, select the Can be start node check box. 4 In this node, we need the AnalysisYear variable as a context variable. Therefore, press the + button to add a context variable, and select the only available Model Variable, AnalysisYear, from the dropdown list. Creating this node is recognized as the first step to build a model. When you create the node, a list with notes and warnings will appear in the Notes and Warnings pane (the lowest pane in the Modeler window). If this pane is not displayed, select Notes and Warnings from the View menu. The Notes and Warnings pane shows the most important steps to be undertaken to arrive at a model that is syntactically correct. Warnings (the yellow warning signs) indicate crucial steps (or important omissions). Notes point the user to possible omissions that are (at this stage of model building) not prohibitive for running the model. However, note that dependent on the stage of model building (the position of nodes in the model, the number of successors, and so on) new warnings or notes may appear and notes may switch to warnings (and vice versa). You can change the colors and shapes of the nodes through the Graphical Options command in the View menu. The SQL Query of the first node 1 Press the Edit button in the SQL query frame of the Edit intermediate node dialog. The Build SQL Query dialog will appear. 2 Now, we need data from two tables: Orders and Order Details. Put both tables in the joiner area (double-click both). 3 First, both tables must be joined by the key-field OrderID. To do this, select the OrderID field of the Orders table, drag it to the Order Details table, and drop it on the OrderID field of the Order Details table. A thin line will be visible between the two tables. In the SQL query, we define two variables: AverageFreightCosts and AverageDiscountPercentage. See the following steps. 10-5

96 Tutorial 4 AverageFreightCosts can be defined as follows: Double-click on the Freight field in the Orders table. This field will now appear in the Selected Fields pane of the Build SQL Query dialog. Enter AverageFreightCosts in the Alias column. Since we want the average, we have to select the AVG function in the Aggregation column of the pane. 5 The calculation of AverageDiscountPercentage is more complicated and must be done through an expression. Select + in the Expressions pane. This will start the Build Expression dialog. Enter AverageDiscountPercentage in the Variable field. Subsequently, enter the following expression in the lowest pane of the Build Expression dialog: 100 * SUM([Order Details].Quantity * [Order Details].UnitPrice * [Order Details].Discount) / SUM([Order Details].Quantity*[Order Details].UnitPrice) NOTE Select OK to save the expression. In this expression, we first calculate the total Discount as an amount (so dependent on the total order value) and then divide this by the total order value. By multiplying this with 100, we get the average discount as a percentage. The average discount percentage is: (Total discount amount / Total order value) * 100 Now, we must make sure that the analysis is done for the as yet unknown calendar year. That is, we must define a condition in which this calendar year is a parameter that will get its value if the Diagnostic Model is run. 6 The value of this calendar year is stored in the intermediate node s context variable AnalysisYear. We can use this context variable as a parameter in the query as follows: Add a new condition: select + on the Conditions pane to start the Build Condition dialog. We will apply the condition to the OrderDate field of the Orders table. Therefore, select this field in the Fields drop-down list. Select = from the Operator drop-down list. In the Value, Expression, Parameter field, we can either enter a value (static condition), an expression (press the ellipsis button) or a parameter. If you use the drop-down list, then you will find the context variable of the intermediate node available for inclusion as a parameter. 10-6

97 Tutorial Select the <[AnalysisYear]> parameter from the drop-down list and press the Insert button. In the lowest pane, you must now manually change the condition to the following: DATEPART(yyyy,dbo.Orders.OrderDate) = <[AnalysisYear]>. We now have a dynamic condition, the value of AnalysisYear being dependent on the value of the corresponding Context Variable in the intermediate node. If our node is used as a starting node, this value will be entered by the end-user prior to running the model (will be shown later in this Tutorial). Select OK to save the expression and to close the Build Condition dialog. 7 The resulting SQL query must be: SELECT AVG(Orders.Freight) AS AverageFreightCosts, 100*SUM([Order Details].Quantity*[Order Details].UnitPrice*[Order Details].Discount)/SUM([Order Details].Quantity*[Order Details].UnitPrice) AS AverageDiscountPercentage FROM dbo.orders AS Orders INNER JOIN dbo.[order Details] AS [Order Details] ON (Orders.OrderID = [Order Details].OrderID) WHERE (DATEPART(yyyy,Orders.OrderDate) = <[AnalysisYear]>) (try to test the query in the Build SQL Query dialog: select! in the Preview SQL Query pane. What happens?) 8 Select OK to leave the Build SQL Query dialog. 9 Select OK to leave the Edit intermediate node dialog. Adding additional Model Variables Before we continue to extend the model, we will have to add variables to the list of Model Variables. The Query results of the first Intermediate node are needed later in the model and, hence, we need model variables to carry these results. 1 Open the Edit Model dialog (select Model from the Edit menu) and add two Model Variables: AverageDiscount and AverageFreightCosts. 10-7

98 Tutorial 2 Both are numeric variables that will be stored as a float when saving the conclusions to a SQL server database. An initial value query is irrelevant for these variables. 3 Select OK to leave the Edit Model dialog. The second intermediate node 1 Insert a second intermediate node in the drawing area. In this node we will calculate the overall average order value (turnover) per employee. 2 Edit the node: enter a Short Title, a Title and a Description. We do not want this node to act as start node, so leave the Can be start node check box empty. 3 In this node, we must add all three Model Variables defined until now as Context Variables. The Model VariableAnalysisYear will be actually used as a parameter in the query that we will construct for this node. The other two will not be used in the query. The only reason that we include them in the context of the present node is that we can transfer them to the next node (where they will be used in the query). The SQL query of the second node To calculate the Average order value (per employee) we must follow a two-step approach. First, we calculate the total order value for each employee in a subquery. Second, we take the average of these individual values in the main query. Take the following steps to create both queries: 1 Select Edit in the Edit intermediate node dialog to start the Build SQL Query dialog. 2 Select Define Sub Query (the S button at the right side of the table list) to create a subquery. This will start the Sub query dialog. This dialog has the same layout as the Build SQL Query dialog, except for the Name field (enter OrderValueByEmployee as a name for the subquery). 3 Put the Orders and the Order Details tables in the joiner area and create a join between the OrderId fields in both tables. 4 Select the EmployeeID field from the Orders table. 10-8

99 Tutorial 5 Build the following expression for the total order value: NOTE TotalOrderValue = SUM(([Order Details].Quantity*[Order Details].UnitPrice* (1-[Order Details].Discount))-Orders.Freight) The []s around Order Details are needed because there is a space in this table name. They will be automatically added if the Tables / Fields drop-down boxes in the Build Expression dialog are used. Since this expression contains a SUM, in the resulting subquery the selected field EmployeeID will be automatically put in a GROUP BY clause. In other words, the total order value (or turnover) is calculated for each employee. 6 As in the query of the first Intermediate node, also here we want to restrict to one calendar year (the AnalysisYear). Hence, also here you must create the following condition. (DATEPART(yyyy,Orders.OrderDate) = <[AnalysisYear]>) 7 Select OK in the Sub Query dialog and return to the main query. 8 Now the subquery is shown in the list of tables. It can be treated as any other table. Put the subquery in the joiner area of the Build SQL Query dialog. 9 Select the TotalOrderValue field (double click on it). Enter AverageOrderValue as alias name and choose the AVG aggregation (that is, take the average). The resulting query should be: SELECT AVG(OrderValueByEmployee.TotalOrderValue) AS AverageOrderValue FROM ( SELECT Orders.EmployeeID AS EmployeeID, SUM(([Order Details].Quantity*[Order Details].UnitPrice*(1-[Order Details].Discount))- Orders.Freight) AS TotalOrderValue FROM dbo.orders AS Orders 10-9

100 Tutorial INNER JOIN dbo.[order Details] AS [Order Details] ON (Orders.OrderID = [Order Details].OrderID) WHERE (DATEPART(yyyy,Orders.OrderDate) = <[AnalysisYear]>) GROUP BY Orders.EmployeeID ) AS OrderValueByEmployee 10 Select OK to leave the Build SQL Query dialog. 11 Select OK to leave the Edit intermediate node dialog. Connecting the first two nodes and editing the Branch 1 Connect the first two nodes through a branch: first select (=click on) the first node (Av. Discount and Freight) and then select the second node (Average Order Value). A dashed helper arrow should now be visible. To actually add the branch, select Branch to Connect Nodes from The Insert menu (or click the Connect Nodes by inserting a Branch button in the toolbar). The branch is now visible as a solid arrow with the text Branch 1. This latter text can be replaced by any other text by editing the branch and specifying a short title (which should cover the purpose of the branch or phrase the branch condition). 2 Start the Edit Branch dialog: double-click the branch, or select the branch and press Enter, or select the branch and subsequently select Edit selected object from the context menu. The Edit Branch dialog can be used to enter some meta data (short title, description), to specify a condition (not applicable to our branch, because it is the only branch that starts in the first intermediate node), and to specify the information flow (mapping). In the mapping area, three boxes are shown (note: they look exactly the same as Tables in the joiner area of the Build SQL Query dialog, but they have no relation with SQL Server tables): The box in the upper left corner lists the context variables of the origin node (that is, the node Av. Discount and Freight): in our example, only the model variable AnalysisYear

101 Tutorial The box in the lower left corner lists the variables returned by the SQL Query of the origin node. In our case, AverageFreightCosts and AverageDiscountPercentage. The box at the right side lists the (required) context variables of the destination node (Average Order Value). In our case: AnalysisYear, AverageDiscount and AverageFreightCosts. You must indicate how the required context variables of the destination node get their values when the model is run. The required context variable AnalysisYear will get its value by copying the value you already assigned in the first node. To realize this, you must map the AnalysisYear context variable of the first node on that of the second node. 3 Map the AnalysisYear context variable of the first node on that of the second node: select AnalysisYear in the upper left box, drag it to the right box and drop it on AnalysisYear in the right box. Now a connection-line will appear. This line is not a join; it only indicates how values are transferred. 4 The required context variables AverageDiscount and AverageFreightCosts were not in the context of the origin node and, hence, they should get their value from the query results of the origin node. Map AverageFreightCosts and AverageDiscountPercentage from the lower left box on AverageFreightCosts and AverageDiscount in the right box respectively

102 Tutorial The Edit Branch dialog should now look like: Figure 53 Branch between Av. Discount and Freight and Average Order Value node 5 Now, the branch is specified completely. Select OK to leave the dialog

103 Tutorial The third intermediate node The third Intermediate node is the node where the figures of individual employees will be compared with the averages calculated in the previous two nodes. After this node, the analysis will continue for those employees that performed below average. In the third intermediate node, we may want to use the average order value calculated in the second node. To carry that information, we need a new model variable. 1 Add the variable TargetOrderValue through the Edit Model dialog. This is a numerical model variable (to be stored as a float in SQL Server). 2 Add a new intermediate node to the drawing area and call it Find bad performers. Edit the new node: take the following steps in the Edit intermediate node dialog. Enter Find bad performers in the Short title field Mark the node as a potential start node (hence, the user can also start the analysis here). Add title and description. Include all available Model Variables in the required context: in the query of this node we will use them all, either to calculate other variables or in the condition. The SQL query of the third node 1 Start the Build SQL Query dialog. Add the Orders, Order Details, and Employees tables to the joiner area. Join Orders and Order Details with the OrderID field. Join Orders and Employees with the EmployeeID field. 2 Define the following five expressions in this query. The difference between the average order value (carried by the model variable TargetOrderValue) and the individual turnover of each employee: ValueBelowTarget = <[TargetOrderValue]> - SUM(([Order Details].Quantity* [Order Details].UnitPrice*(1-[Order Details].Discount))-Orders.Freight) NOTE The context variables of the node are available as parameters (<[ ]>) for the query in the Build Expression dialog

104 Tutorial The employee name. This is a concatenation of the employee s title of courtesy, their first name, last name, and job title (do not forget the spaces): EmployeeName = Employees.TitleOfCourtesy + ' ' + Employees.FirstName + ' ' + Employees.LastName + ' (' + Employees.Title + ')' This is the only expression without an aggregation function and will, therefore, (automatically) be included in the GROUP BY clause of the query. The employee experience. This is the average number of months the employee is at the service of the company. The average is taken for all order dates in the year of analysis. Employee (lack of) experience will later in the analysis be used as one of the possible explanations for a below-average (or below target) individual turnover: EmployeeExperience = AVG(DATEDIFF(mm,Employees.HireDate,Orders.OrderDate)) The ratio of the average discount percentage offered by the employee and the overall discount percentage. This ratio will later in the analysis be used as one of the possible explanations for a below-average turnover. The higher the ratio, the higher the discounts offered by the employee. DiscountRatio = (100*SUM([Order Details].Quantity * [Order Details].UnitPrice*[Order Details].Discount) / SUM([Order Details].Quantity * [Order Details].UnitPrice)) /<[AverageDiscount]> The ratio between the average freight costs related to the delivery of the employee s orders and the overall average freight costs: FreightCostsRatio = AVG(Orders.Freight)/<[AverageFreightCosts]> Also this ratio is used as a possible explanation for a below-average turnover. The higher, the more expensive were the deliveries

105 Tutorial 3 The resulting SQL query is: SELECT <[TargetOrderValue]> - SUM(([Order Details].Quantity*[Order Details].UnitPrice*(1-[Order Details].Discount))-Orders.Freight) AS ValueBelowTarget, Employees.TitleOfCourtesy + ' ' + Employees.FirstName + ' ' + Employees.LastName + ' (' + Employees.Title + ')' AS EmployeeName, AVG(DATEDIFF(mm,Employees.HireDate,Orders.OrderDate)) AS EmployeeExperience, (100*SUM([Order Details].Quantity*[Order Details].UnitPrice*[Order Details].Discount)/SUM([Order Details].Quantity*[Order Details].UnitPrice))/<[AverageDiscount]> AS DiscountRatio, AVG(Orders.Freight)/<[AverageFreightCosts]> AS FreightCostsRatio FROM dbo.orders AS Orders INNER JOIN dbo.[order Details] AS [Order Details] ON (Orders.OrderID = [Order Details].OrderID) INNER JOIN dbo.employees AS Employees ON (Orders.EmployeeID = Employees.EmployeeID) WHERE (DATEPART(yyyy,Orders.OrderDate) = <[AnalysisYear]>) GROUP BY Employees.TitleOfCourtesy + ' ' + Employees.FirstName + ' ' + Employees.LastName + ' (' + Employees.Title + ')' 4 Select OK to leave the Build SQL Query dialog

106 Tutorial ValueBelowTarget as Priority field The query defined in the third Intermediate node will render a recordset with a record for each employee. One of the principles of the Diagnostic Analyzer is that with each of these records we will visit the next node of the model. In this respect, there are two possibilities: Each of these records has an equal weight (record weight = 1 / nr. of records) and the result nodes that will be created as a result of the visit to the next model node have the same weight (result node weight = record weight * weight origin result node). The weight of a record is determined by the value of the Priority field. The next model node will only be visited for those records that have a Priority field above a certain threshold (and at least > 0). This latter is the required situation for our node. The weight of a record is dependent on the value of ValueBelowTarget: The higher this value, the more important the employee contributes to the problem of sub-optimal turnover and the more important the conclusions we will formulate for the employee. Therefore, you must perform the following actions in the Edit intermediate node dialog (for the third node): 1 Select the Based on Priority field radio button in the Priority value frame 2 Select the ValueBelowTarget field from the drop-down list. 3 You can leave 0 as a threshold. If you now view the query, you will see that an ORDER BY clause is added: ORDER BY ValueBelowTarget DESC 10-16

107 Tutorial The Edit intermediate node dialog should now look like: Figure 54 Settings for third intermediate node 4 Select OK to close the Edit intermediate node dialog 10-17

108 Tutorial Connecting the second and the third node 1 Connect the second and third node through a branch. 2 Edit the branch: enter Short Title, for example Calculate individual performance, and Description. 3 Draw the mappings. The mapping of the context variables of the source node to the required context variables of the destination node is straightforward: Map the AverageOrderValue variable (field in the record set of the query of the source node, displayed in the lower left box of the mapping area) on the TargetOrderValue context variable of the destination node. Map the AnalysisYear, AverageDiscount and AverageFreightCosts variables of the source node (displayed in the upper left box of the mapping area) on the corresponding context variables of the destination node. The fourth Intermediate node Now we are entering the decision-area of the model. Based on the calculations in this node, we will decide whether the below-average turnover is caused by lack of experience of the employee or whether we will have to continue the search. To carry the information from the previous node to the current and future nodes, we again need additional model variables. Add the following variables through the Edit Model dialog: EmployeeName (String variable, to be stored as a varchar with length 100), DiscountRatio (Numeric variable, to be stored as a float) and FreightCostRatio (Numeric, float). 1 Add a new intermediate node. Call it Employee Experience. 2 Include the following Model variables in the required context: AnalysisYear, EmployeeName, EmployeeExperience, DiscountRatio and FreightCostRatio. As argued, the calculations in this node are used for a decision: conclude or continue. The decision will be based on the value of the context variable EmployeeExperience. However, only variables (fields) returned by a SQL query can be used for branching decisions. Therefore, we must build a query, which does nothing but returning the value of the EmployeeExperience context variable

109 Tutorial 3 Start the Build SQL Query dialog for the new node. Subsequently, select + in the Expressions pane to start the Build Expression dialog. Create the following expression: EmployeeExperience = <[EmployeeExperience]> NOTE (so, no tables in the joiner). 4 Save the query and the node. 5 Connect the third node with the new node and draw the mappings in the Edit Branch dialog. The node specifications are not yet ready, but we will first add the remaining nodes. The remaining nodes 1 Add the first Conclusive node. Double-click the conclusive node to start the Edit conclusive node dialog and enter Little Experience in the Short title field. 2 Enter the text Employee in the Start conclusion text with text box. 3 Include the following model variables as context variables: EmployeeName. Put the text has not much working experience ( in the second cell (Connecting text). EmployeeExperience. Put the text months). Year of analysis: in the second cell. AnalysisYear. Put. In the second cell. Now the conclusion text should read like: Employee <EmployeeName> has not much working experience (<EmployeeExperience> months). Year of analysis: <AnalysisYear>. The placeholders < > will at run time be replaced by real values. 4 Add the fifth intermediate node; call it Average Discount. 5 Include the model variables AnalysisYear, DiscountRatio, EmployeeName and FreightCostRatio in the context. 6 Build the following query in this fifth node (use the Build Expression dialog in the Build SQL Query dialog): SELECT <[DiscountRatio]> AS DiscountRatio 10-19

110 Tutorial 7 Connect the fourth intermediate node (Employee experience) both with the first conclusive node (Little Experience) and the fifth intermediate node (Average Discounts). Draw the mappings in the branches. 8 Now reopen the fourth intermediate node (Employee experience). In the Branch field frame of the dialog you can now select a branch field (this frame is disabled as long as no or only one branch leaves the node). Select the EmployeeExperience field as the branch field and set the data type to Float. You can now use this branch field in the emanating branches to define a condition. 9 Edit the branch connecting the fourth Intermediate node with the Conclusive node. In the Branch Condition frame, select the Value Range option. Put the value 30 in the rightmost text box of the frame and select < from the dropdown list left to that text box

111 Tutorial The Edit Branch dialog should now look like: Figure 55 Branch settings NOTE In the source node, EmployeeExperience is both available as a context variable and as a field in the result set. Since they hold the same value, it does not matter which of the two is mapped onto the context of the destination node

112 Tutorial 10 Also edit the branch connecting the fourth and the fifth Intermediate node. Define the condition EmployeeExperience >= 30 (put 30 in the rightmost text-box of the Branch Condition frame and select <= from the drop-down list right to that text box). 11 Add the second Conclusive node; call it Too high discounts. Include the model variables EmployeeName and AnalysisYear in the list of context variables. Make sure that the following conclusion text appears in the lower pane: Employee <EmployeeName> offered too high discounts to his / her customers. Year of analysis:<analysisyear>. 12 Add a sixth intermediate node: Call it Average Freight Include the model variables AnalysisYear, EmployeeName and FreightCostRatio in the context. Use the Build Expression dialog in the Build SQL Query dialog to create the following SQL query SELECT <[FreightCostRatio]> AS FreightCostRatio 13 Connect the fifth intermediate node (Average Discount) to the second conclusive node (Too High Discounts) and to the sixth intermediate node (Average Freight). 14 Reopen the fifth intermediate node and select DiscountRatio as the Branch field (the field on the basis of which we decide to continue our analysis). Set the data type to Float. 15 Edit the branch connecting the fifth intermediate node (Average Discount) to the second Conclusive node (Too high discounts). Set the Branch condition to: DiscountRatio > 1 16 Edit the branch connecting the fifth intermediate node (Average Discount) to the sixth intermediate node (Average Freight) and set the condition to: DiscountRatio <=

113 Tutorial 17 Add a third conclusive node, call it Too High Freight Costs and configure it so that the conclusion text becomes: The company paid too much freight costs for the sales orders of employee <EmployeeName>. Year of analysis:<analysisyear>. 18 Add the last (fourth) conclusive node. This node will be visited if none of the other conclusive nodes was visited (by an employee with a less than average turnover). Call it No Explanation and make sure that the conclusion text reads like: There is no explanation for the below-average turnover of employee <EmployeeName>. Year of analysis <AnalysisYear>. 19 Connect the sixth intermediate node (Average Freight) to the third and the fourth conclusive node. Draw the mapping in the branches. 20 Reopen the sixth intermediate node and select FreightCostRatio as branch field (data type float). 21 Set the condition for the branch connecting the sixth intermediate node (Average Freight) with the third conclusive node (Too High Freight Costs) to: FreightCostRatio > 1 22 Set the condition for the branch connecting the sixth intermediate node (Average Freight) with the fourth conclusive node (No explanation) to: FreightCostRatio <= 1 Prepare for execution Compare your model with the sample Northwind model (Northwind.dam file) that is delivered with the ibaan Diagnostic Analyzer software. Before executing the model, you must specify some default report settings: 1 select Default report settings from the Edit menu to start the Default Report Settings dialog. The model contains two possible start nodes: Av. Discount and Freight and Find Bad Performers. You must select one of these as the actual start node for the execution of the model

114 Tutorial NOTE 2 Choose the first intermediate node (Av. Discount and Freight) from the dropdown list. If you select a start node, the context variables of that node are displayed in the Assign initial values to context variables grid. The Av. Discount and Freight node contains only one context variable, AnalysisYear, to which you must assign a value. The Find Bad Performers node contains four context variables: AnalysisYear, TargetOrderValue, AverageDiscount and AverageFreightCosts. 3 Enter 1996 as initial value for the AnalysisYear context variable. Also try the ellipsis ( ) button in the Initial Value column. 4 Enter the remaining default report settings based on the following screenshot: Figure 56 Default Report settings 5 Select OK to close the dialog. If you completed all steps, there should be no warning left in the Notes and Warnings pane of the Diagnostic Analyzer Modeler dialog. Now you can execute the model: 10-24

115 Tutorial Select Run Model from the File menu. Or: Select Execute the Model (exclamation mark button) from the toolbar. Or: Press F9. The report that pops up after finishing the run should look like: Figure 57 Sample conclusions report You can use the Tree Analyzer application to investigate the Result tree in detail. To start this application, select Tree Analyzer from the ibaan Diagnostic Analyzer menu. Subsequently, press the Open button on the Tree Analyzer s Control Panel, and select the Result Tree file. For example: C:\temp\Northwind.drr. The name and location of this file is specified in the Result Tree frame in the Default Report Settings dialog

116 Tutorial See the following figure for a sample Result tree. Figure 58 Sample Result tree 10-26