Predictive Analytics Workshop. Workbook Public Sector Edition

Transcription

1 Predictive Analytics Workshop Workbook Public Sector Edition

2

3 Contents 1. NAVIGATING IBM SPSS MODELER PREDICTIVE IN 20 MINUTES SEGMENTATION MODELING ENTITY ANALYTICS CLASSIFICATION MODELING TEXT ANALYTICS DEPLOYMENT Contents Page 1

4 Overview This hands-on IBM SPSS Data Mining Workshop is an instructor-led session using IBM s data mining and predictive modeling software and is designed for those who are familiar with predictive analytics. Through this workshop you will experience firsthand how IBM SPSS Modeler works and how easy it is to implement predictive analytics. Page 2

5 1. Navigating IBM SPSS Modeler 1.1 Start Modeler if it s not already open. 1.2 Review the various palettes near the bottom of the screen. The names of the nodes reflect the operation or function accomplished with that node. 1.3 From the Sources palette, select the Database node and add it to the canvas. This can be done by the following methods: Appendix Page 3

6 Double-click the node on the palette. Drag and drop the node onto the canvas. Page 4

7 1.4 From the Output palette, select the Table node and add it to the canvas. Then join it to the Database node using any of the following methods: Double-click on the node in the pallette to automatically add it to the stream and join it to the selected node. Drag and drop the node from out of the pallette and on to the canvas. Select the first node, right click and select Connect from the context menu, and then left click on the second node to connect it. Click and hold the middle mouse button on the first node, move the cursor to the second node and release when the cursor is ontop of the second node. 1.5 To delete the connection, right click on the joining arrow and select Delete Connection. The other methods listed in 1.4 can also be used to do this. Delete the connection. 1.6 To remove a node, select it and press the Delete button, or right click on the node and select Delete from the context menu. 1.7 From the File menu, select Close All Streams. Appendix Page 5

8 2. Predictive in 20 Minutes 2.1 Start a new stream by clicking on File > New stream. 2.2 From the Sources palette, add an Excel node. Page 6

9 2.3 Double-click the Excel node to open a dialog box. Use the Data tab to import the Patient Data.xlsx file from C:\PS_ModelerWorkshop\PredictiveIn20Minutes... directory. Be sure you specify the File Type as Excel 2007, 2010 (*.xlsx). 2.4 Click the Preview button at the top of the dialog box to see the first 10 records in the file. This is an extraction of data from a clinical trial studying the impact of various behaviors and demographics and conditions on the risk of experiencing a heart attack. Click OK to close the preview. Appendix Page 7

10 2.5 From the Field Ops palette, add a Type node to the canvas and connect it to the data source. Page 8

11 2.6 Open the Type node and click the Read Values button to scan the data as well as to display and update the range of values. Modify the Role of the heartattack_s field to Target. Click OK. 2.7 From the Graphs palette, add a Distribution node and connect it to the Type node. Edit the Distribution node so that the heartattack_s field will be graphed. Click OK (or Run to execute the stream now). Appendix Page 9

12 2.8 Run the stream to produce the graph (if not already done). To do this, right click on the node and select Run from the context menu, or click the Run button in the toolbar near the top of the screen. 2.9 Looking at the output, we see that of the 7,998 patients in this dataset, an alarming 43.85% have reported a history of heart attack while the remaining 4,491 patients have not. The task now is to build a model to understand the relationships within the data that indicate greater risk for heart attack. A C&R Tree is one modeling method to do this From the Modeling palette, add the C&R Tree node to the Type node. Note that the C&R Tree node name changes to heartattack_s when it is connected to the Type node. This is because we defined that variable s role as the target in Step 2.6, and the modeling node honors that setting and automatically uses that variable as the target. Page 10

13 2.11 Open the C&R Tree node to view the settings before running the model. In this example we will use the default settings and make no changes, though if desired, you have the capability to manage the settings. Click Run The C&R Tree model is generated, automatically added to the canvas and named heartattack_s. Appendix Page 11

14 2.13 Double-click on the generated model to review the output. The Model tab contains predictor importance and rules for the model (not shown). Click the Viewer tab for a visual depiction of the tree Looking at the tree diagram, we see that Smoking is the best predictor of heart attack, and that a greater concentration of those who have experienced a heart attack (77% of the cases in the data) are smokers, older than 60.5 years of age. But among non-smokers, there s a greater concentration of those who have had heart attacks among patients with of diabetes (81.3%). Using Decision Tree analyses enables us to uncover not only the best predictors of heart attack or student risk, or other targets specific to our field but also the relationships or interactions between the variables which we would otherwise have to define if using other methods. Page 12

15 2.15 Click OK in the Tree Viewer to close it (not shown). Then, from the Output palette, connect a Table node to the generated model and run it Look at the target variable, heartattack_s, and last two columns in the table. The variable, heartattack_s records observed outcome - whether the patient has had a history of heart attack or not and the second to last column is the prediction from the model. The last column is the confidence in the prediction. Appendix Page 13

16 2.17 Again from the Output palette, connect an Analysis node to the generated model and run it. This shows the overall accuracy of the model If desired, save the stream to a working directory before closing it (File > Save stream as... and name it as desired) Click File... Close all streams. Page 14

17 3. Segmentation Modeling 3.1 Open SegmentationModeling.str from the workshop directory. Go to Menu, select File, select Open Stream and locate C:\PS_ModelerWorkshop\SegmentationModeling\SegmentationModeling.str or double-click the file. Appendix Page 15

18 3.2 Attach a Table node from the Output Palette to the Type node, and run it. Review the data about alumni who have contributed donations to the university. Using the red X in the Table viewer, delete the Table view when done. 3.3 Run the Data Audit node, named 8 Fields and view the distributions of the data fields and summary statistics. Double-click one of the graphs to see a more detailed version. Again, use the red X to delete the output when done. Page 16

19 3.4 Open the Type node and change the Role for the amount_sum field to Target. Only the variables defined as Input fields will be used to determine the clusters. Click OK. 3.5 From the Modeling palette, add the K-Means Cluster node to the Type node and double-click on it to edit it. Change the Number of Clusters setting in the Model Tab to 5 (if necessary) and then run it. Appendix Page 17

20 3.6 Double-click the generated model to view the results of the clustering method. The output is divided into two viewers: on the left is a model summary; on the right, Cluster Sizes. From the right viewer s drop-down box, select Predictor Importance. 3.7 The right viewer displays a graph with the fields ranked in order of importance for cluster creation. From the left viewer s drop-down box, select Clusters. Page 18

21 The left viewer displays a grid with the clusters listed from left to right in order of size largest to smallest and the fields listed in descending order based on importance. Click one of the cells to view how a cluster distribution compares to the remaining records in the right viewer. 3.8 Select the heading for the left-most cluster, hold down the Shift button and click on the right-most cluster to select the entire table. This will change the right viewer to the Cluster Comparison view, illustrating how the clusters are different based on the fields used in the models. How are the clusters different from each other? Do you see particular donor types emerging through or defined by the clusters? Appendix Page 19

22 3.9 Click OK to close the cluster viewer From the Field Ops palette in the Modeler Workbench, select a Reclassify node and attach it to the generated model Double-click on the Reclassify node to edit it. From the variable drop-down list, choose the variable $KM- K-Means, and in the New field name field, enter Clusters and then the Get button. Based on your review of the cluster descriptions, assign names or labels to the clusters which are more descriptive of the typical cluster member. Click OK when done. Page 20

23 3.12 Add a Table node to the generated model to see the cluster membership variable added to the dataset (not shown) Save the Stream to a local directory before closing it. Appendix Page 21

24 3.14 Appendix: Feature Selection and Predicting Donation Amount Open the stream labeled Donor Clusters and Predicted Giving SOLUTION.str located in C:\PS_ModelerWorkshop\SegmentationModeling. Feature Selection: Data mining problems may involve hundreds, or even thousands, of fields that can potentially be used as inputs. As a result, a great deal of time and effort may be spent examining which fields or variables to include in the model. To narrow down the choices, the Feature Selection node can be used to identify the fields that are most important for a given analysis. Feature selection consists of three steps: Screening. Removes unimportant and problematic inputs and records, or cases such as input fields with too many missing values or with too much or too little variation to be useful. Ranking. Sorts remaining inputs and assigns ranks based on importance. Selecting. Identifies the subset of features to use in subsequent models for example, by preserving only the most important inputs and filtering or excluding all others. In an age where many organizations are overloaded with too much data, the benefits of feature selection in simplifying and speeding the modeling process can be substantial. By focusing attention quickly on the fields that matter most, you can reduce the amount of computation required; more easily locate small but important relationships that might otherwise be overlooked; and, ultimately, obtain simpler, more accurate, and more easily explainable models. By reducing the number of fields used in the model, you may find that you can reduce scoring times as well as the amount of data collected in future iterations. Finally, and perhaps most importantly, remember that an essential component of Predictive Analytics is action taken on the insights gained from analyses; Predictive Analytics is nothing without action. But when action is taken, by definition the data necessarily change. When the data change, the models become obsolete; some predictors may deflate in their importance and be replaced by others which have gained in Page 22

25 their importance. The feature Select node is an essential component in the modeling stream where data is dynamic and constantly maturing. In this way, the Feature Selection node will update the choice of variables as your data matures. If you are using an automated modeling node (discussed in the section on Classification modeling), then together with the Feature Selection node you have the capability to create models which will update themselves in the wake of their own predictions Double-click on the Feature Selection model to view the list of variables determined to be important contributors to building the model for predicting donation amounts Click OK to close this dialog box after reviewing the list of variables which will be included in the modeling process. Appendix Page 23

26 3.17 Double-click on the C&R Tree modeling node and move the variable $KM-K-Means out of the Inputs field (if necessary). This is because the variable Clusters contains the very same information We will not make any changes to the dialog other than this, but if you wanted to change any of the build options for the C&R Tree analysis, you can access those parameters from the Build Options and/or the Model Options tabs. Click Run to execute the analysis Double-click on the model to view the results, and select the View tab. Page 24

27 In the resulting tree diagram above, age is determined to be the best predictor of giving, with cluster membership being the next best predictor, followed by whether or not a prospective contributor attended an alumni event. So, following the branch to the left, if a contributor is less than or equal to 70.5 years old, and belongs to the cluster labeled Elise Keaton she would be predicted to donate on average $1,132, whereas if someone is not in that cluster, we d predict an average donation of $224. Following the tree further along that branch (not shown) we see that event attendance is the next best predictor; if one did not attend an alumni event the predicted donation drops to an average of $145. Click on OK to close the viewer Right-click on the Graph node labeled Predicted contribution and run it. The cluster labeled Mike Brady is predicted to contribute the most, both in total amount predicted as shown, and also by average donation (not shown). Those in the cluster labeled Charlie Harper are likely not very good prospects for soliciting donations, as they are predicted to contribute the least both in total and on average Click OK to close the graph, and then select File > Close stream... to close the stream. Do not save any changes to this stream. Appendix Page 25

28 4. Entity Analytics IBM SPSS Modeler Entity Analytics adds a completely new dimension to IBM SPSS Modeler predictive analytics. Whereas predictive analytics attempts to predict future behavior from past data, entity analytics focuses on improving the coherence and consistency of current data by resolving identity conflicts within the records themselves. An identity can be that of an individual, an organization, an object, or any other entity for which ambiguity might exist. Identity resolution can be vital in a number of fields, including customer relationship management, fraud detection, anti-money laundering, and national and international security. 4.1 Open the Utilities Risk Demo.str in C:\PS_ModelerWorkshop\Entity Analytics. Page 26

29 4.2 From the Export palette, drag an EA Export node on to the canvas and connect it with the Type node at the top of the stream. Once connected, double-click on the EA Export node to open it. Appendix Page 27

30 4.3 Entity Analytics requires a repository within which entities actors, vessels, etc. can be compared on a number of features and identified as being the same or different entities. The repository has to be configured first. Click the drop-down arrow in the Entity Repository field, and select Browse. 4.4 In the Entity Repositories dialog, click on the drop-down arrow for the field labeled Repository Name and select <Create/Add New Repository> Page 28

31 4.5 In the resulting dialog, labeled Create/Add repository: Step 1 of 2 (shown below) enter the New Username administrator and for the New Password enter IBMDem0s. When done, click on Next. In Step 2 of 2, enter PAW as the New Repository Name, and then click OK. 4.6 In the Entity Repository dialog, click OK again (not shown). Appendix Page 29

32 4.7 In the EA Export dialog, click on the drop-down arrow for the Source tag field and select Add New Source Tag. This will raise another dialog; enter PAW as the new source tag and click OK. Page 30

33 4.8 After the source tag has been successfully added, click on the drop-down arrow for Entity Type and select PERSON. 4.9 Next, in the variable drop-down list next to the Unique Key field, select the variable key Appendix Page 31

34 4.10 At this point, we have to map the variables in the data to the appropriate features in the EA repository. This can take some time to do. However, once done, you can export the mapping definitions as an XML file and reuse them. This has already been done for you for this workshop. So, instead of mapping each and every individual variable to the appropriate feature in the EA repository, we ll use the previously created XML file which already has the mapping definitions. Click on the Import Mapping pushbutton at the lower left side of the dialog (not shown). In the Open dialog, change the Look in directory to C:\PS_ModelerWorkshop\Entity Analytics and select the DMWmapping.xml file, and then click on Open. Once the mapping has been imported successfully, select OK to close that alert box. Next, map the dob field to DOB.DATE and the dod field to DOD.DATE. Once this is complete, the red text will be replaced by the correct feature name. Page 32

35 4.11 After the feature mapping is complete, click on Run to execute the entity resolution engine When the Execution Feedback is complete, the repository is built and is ready to use as a data source. So, from the Sources tab, drag over an Entity Analytics source node, place it on the canvas underneath the square EA Repository node, and connect it to the Filter node already in the stream (as shown below). Appendix Page 33

36 4.13 Double-click on the EA source node to open it. From the Entity Repository drop-down list, select Click the check box in the Include column for the DMW Data Source. The number of records should report 850; if it does not,click the push button at the top of the dialog to Refresh. Finally click OK This example begins by both 1) building a model from historical data and using it to predict risk of default in applications for new service, and 2) building the entity repository. Right-click on the distribution node at the upper right end of the stream and select Run. This generates the following distribution graph, which displays the number of new applicants for service both at risk of default and those who are not. Page 34

37 From the graph we see that 54 new applicants for service have been scored as being at risk for potential default. In other words, 36% of the new applicants could likely benefit from alternative payment or billing options from the utility company, rather than receive standard service. Yet, how many of the new applicants are actually return customers who have already defaulted on their services and are fraudulently applying for new accounts under a different identity? Using the lower portion of the stream in this example the one in which we use the Entity Repository we can determine who from among the new applicants are likely the same as some old customers who have already defaulted on their services. Now, right-click on the distribution node at the lower right of the stream and select Run. This generates the distribution graph below (shown in part) which displays the increase in the number of potential cases at risk of defaulting on their service. These additional cases may not have been detected as potentially at risk using the model, but after resolving entities, these cases were considered as the same actors who are currently determined as at risk. Appendix Page 35

38 5. Classification Modeling 5.1 Before beginning, click on File... Close all streams (if necessary). 5.2 Open the ClassificationModeling.str stream from the workshop directory. Go to Menu, select File, select Open Stream and locate C:\PS_ModelerWorkshop\ClassificationModeling\ClassificationModeling.str or double-click the file. 5.3 The stream already consistes of a Source node, a Type node, a Data Audit node and a Distribution graph node. The Source node retrieves an SPSS data file containing information about first year students at a university, information like ACT test scores, demographics, etc. The Type node defines the measurement levels and roles of the variables; notice that Student ID is defined as a record identifier, and persist is defined as the target. 5.4 Right-click on the Data Audit node and run it. Again, the results (below) display the summaries of the variables, and the graphs also display the concentration of the target variable relative to each predictor. Page 36

39 5.5 Click the red X to close and delete the Data Audit output (not shown). Next, right-click and run the Distribution node at the top, labeled persist. In the resulting frequency graph, you can see that the target variable is not evenly distributed. So, building models without evening out the data weighting it runs the risk of creating models which will always predict to the mode (or the most frequently represented category, which here is 1). Use this output from IBM SPSS Modeler to automatically calculate the weighting scheme and create a Balance node. As is highlighted above, from the menu at the top of the graph output item, click on Generate > Balance node (boost). Then click OK to close the chart. On the canvas, you ll see a new balance node labeled generated. Drag that node and place it after the Type node and then connect the two (as shown below). Appendix Page 37

40 5.6 Double-click on the Balance node to review the weighting values now in place. Click OK to close. 5.7 We re ready to build the classification model. For this we ll use the Auto Classifier node. The Auto Classifier node estimates and compares models for either nominal (set) or binary (yes/no) targets, using a number of different methods, allowing you to try out a variety of approaches in a single modeling run. You can select the algorithms to use, and experiment with multiple combinations of options. The node explores every possible combination of options, ranks each candidate model based on the measure you specify, and saves the best models for use in scoring or further analysis. From the Modeling palette, drag and place the Auto Classifier node to the canvas and connect it to the balance node. Then double-click on it to open it, and select the Expert tab. Those models which have been checked will be used. Some were not checked because the data did not support their use. You can check or uncheck modeling techniques as desired. We will make no changes to this dialog, but as stated above you have several options and settings to choose from when using this automated modeling node. Page 38

41 Click Run. Then, once the model has been been generated, double-click on it and review the results. As seen below, Modeler retains the three models which perform the best and presents them in ensemble. Ensemble modeling, which happens automatically in the Auto Classifier, combines two or more model nuggets to obtain more accurate predictions than can be gained from any of the individual models. By combining predictions from multiple models, limitations in individual models may be avoided, resulting in a higher overall accuracy. Models combined in this manner typically perform at least as well as the best of the individual models and often better. Above we see that the overall accuracy of the C5 model, which is a decision tree modeling technique, is far better than that of the C&R Tree and Quest models. If we wanted, we could simply uncheck those two and just use the C5 engine. But for now, we ll make no changes. Appendix Page 39

42 5.8 As highlighted above, double-click on the C5 model. In the model tab you can review the rule sets which were generated by the model. A Rule Set tries to make predictions for individual records. Rule sets are derived from decision trees and, in a way, represent a simplified or distilled version of the information found in the decision tree. Rule sets can often retain most of the important information from a full decision tree but with a less complex model. For example, in the highlighted area above, if a student is African-American, has had no contact with the school prior to enrollment, scored between in the reading section of the ACT and graduated from a high school class size of 216 or smaller, then we would classify that student as likely to persist. Click OK to close the C5 model results, and then OK again to close the model results from the Auto Classifier. 5.9 From the Graphs palette, place an Evaluation graph node onto the canvas after the model, and attach it to the model. Then, from the Output palette, place an Analysis node onto the canvas after the model as well, and attach it to the model too. Page 40

43 5.10 Run both the Evaluation and the Analysis nodes. Looking at the evaluation chart first, we generally interpret the rising blue line as indicating the classification accuracy of the model. By default, we selected a Gains chart. Cumulative gains charts always start at 0% and end at 100% as you go from left to right. For a good model, the blue line in the gains chart will rise steeply toward 100% and then level off. A model that provides no information will follow the diagonal from lower left to upper right (shown in the chart as a red line). We can see a steadily increasing classification accuracy (or gain) as the proportion of cases targeted increases. Curiously, the gains level off between the 45 th and 50 th percentile. Still, this chart tells us that if we were to have done no analytics at all, then if we selected 50% of the students in the data at random, likely we d see only 50% of the target among that sample. But with Predictive Analytics, you can now more effectively target 50% of your students and capture 78% of those likely to persist. Looking at the output from the Analysis node below, we see that overall, the model has a 73.55% accuracy. Close the output and then save (if desired) and close the stream. Last, select File... close all streams. Appendix Page 41

44 6. Text Analytics 6.1 Open textanalytics.str from the workshop directory. Go to Menu, select File, select Open Stream and locate C:\PS_ModelerWorkshop\TextAnalytics\textanalytics.str or double-click the file. 6.2 Attach a Table node from the Output Palette to the Excel source node, and run it. Review the data containing officer narratives. Page 42

45 6.3 Right click on the Text Analytics node and select RUN; this begins the extraction process and opens the interactive text analytics workbench. Click Build and, if prompted, Build Now. Use the defaults for this example. 6.4 This populates the categories in the top left section. They can be expanded to view the concepts and descriptors that make them up. Appendix Page 43

46 6.5 Scroll to locate the street category. Left click on it and drag it onto the location category. Releasing it combines the two categories together. 6.6 The street category is combined with the location category and the other subcategories. Page 44

47 6.7 Select weapon and click Display to view the individual comments with occurrences in the bottom right corner and co-occurrence other categories in the upper right corner. 6.8 Select Resource Editor from the drop-down menu in the top right corner. This is where the linguistic resources the libraries and dictionaries - are located. These resources can be fully manipulated for how you want concepts and sentiments handled in the program. Appendix Page 45

48 6.9 Toggle back to the Categories and Concepts view. From the Generate menu, select Generate Model. Close the Text Analytics Workbench and, when prompted, select Update then OK, twice. This returns you to the main screen From the Models pane in the top right, drag the generated model to the palette and connect it to the Excel source node. Page 46

49 6.11 From the Output palette, add a Table node to the modeling node and RUN the Table to see the comment categories Scroll right to view the new categories created. A value of T is placed in a category if the comment contained information which qualifies for inclusion in that category; an F indicates that the text contained no information concepts and/or sentiment which would be categorized into that category. Appendix Page 47

50 6.13 From the Field Ops palette, add a Type node to the model. Double-click on the Type node and select Read Values (not shown). The newly generated categories from the text analysis can now be defined and used as additional inputs into a future predictive policing model Click OK to close the Type node. At this point you have effectively structured what was previously unstructured data. Again, these new insights contained in the text data can now be leveraged together with additional information crime reports, weather patterns, demographic data, events data, etc. for more effective policing, whether the goal is for force deployment, lead generation, traffic safety or anticipating specific calls for service Close all streams now by selecting File.. Close all streams. (No need to save the stream). Page 48

51 6. Text Link Analysis (Optional) Text Link Analysis (TLA) adds a pattern-matching technology which can identify relationships between the concepts in the text data based on known patterns. These relationships can describe which entities are associating with each other, such as the relationships between genes or pharmaceutical agents, whether a car described in the police narrative was used in committing the crime or not, etc. You can use the TLA pattern rules inside the resource templates that are shipped with IBM SPSS Modeler Text Analytics, or you can create/edit your own. Whenever a TLA pattern rule matches the text, it can be exacted as a TLA result and restructured as output data Open TAsolution.str from the workshop directory. From the File menu, select Open Stream and locate C:\PS_ModelerWorkshop\TextAnalytics\TAsolution.str or double-click the file We need to confirm that the appropriate resource template has been loaded for this TLA example. Double click on the TLA node, and then click on the Load button (shown in part below). Appendix Page 49

52 The dialog already reports that the CPP Resources2 (English) template has been loaded and is in use. However, to be sure we clicked on Load. In the resulting dialog, click Import (not shown) and navigate to the working directory, C:\PS_ModelerWorkshop\TextAnalytics, where you ll find the file named CPP Resources2 (English).lrt. Double-click on that file (or select it and then click Open), and if prompted, select OK to overwrite the existing file. Select OK again to return to the Modeler canvas At this point, IBM SPSS Modeler is now able to extract information such as a vehicle description as well as determine automatically if the car is the getaway car or just a squad or witness car. Right-click on the Table node in the TAsolution stream and select Run. This will raise the table view below. In the table view, scroll all the way over to the right to view the results. Page 50

53 6.19 From the Field ops pallet, select a Type node and attach it to the Merge node. Double-click on the Type node to edit it to set the Measurement of the variable SHIFT to Nominal. Be sure to also select Read values and then OK Last, we want to leverage the results from the previous Text Analysis exercise to determine when from within which shift - crimes committed with a firearm occurred most frequently. From the Graphs pallet in the Modeler workbench, select a Graphboard node and attach it to the existing type node. Appendix Page 51

54 6.21 Double-click the Graphboard node to edit it, and select the variable Shift and the thumbnail for Bar of Counts (if necessary) In the Detailed tab, overlay the color with the text analysis variable Category_weapons/gun (shown in part, below) and then click Run. Page 52

55 6.23 The resulting stacked bar chart shows us that the majority of the incidents reported where a firearm was referenced or described in the police narratives occurred during the third shift. While this is a relatively straight-forward illustration of how Text Analytics can be used together with structured data for gaining better insights into your data, it is not difficult to envision how it can also be used for more proactive Predictive Analytics. The final section in this workshop will discuss how Predictive Analytics can be deployed, using the example of Predictive Policing. Appendix Page 53

56 7. Deployment 7.1 Navigate to C:Ps_Workshop\Deployment\CPPdeployment double-click on the file, Commander Dashboard.htm to open it into a browser. 7.2 Browse through some of the content with the presenter, including the force deployment examples and the completed Text Analytics Report. Page 54

57 Notices This information was developed for products and services offered in the U.S.A. IBM may not offer the products, services, or features discussed in this document in other countries. Consult your local IBM representative for information on the products and services currently available in your area. Any reference to an IBM product, program, or service is not intended to state or imply that only that IBM product, program, or service may be used. Any functionally equivalent product, program, or service that does not infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to evaluate and verify the operation of any non-ibm product, program, or service. IBM may have patents or pending patent applications covering subject matter described in this document. The furnishing of this document does not grant you any license to these patents. You can send license inquiries, in writing, to: IBM Director of Licensing IBM Corporation North Castle Drive Armonk, NY U.S.A. For license inquiries regarding double-byte (DBCS) information, contact the IBM Intellectual Property Department in your country or send inquiries, in writing, to: IBM World Trade Asia Corporation Licensing 2-31 Roppongi 3-chome, Minato-ku Tokyo , Japan The following paragraph does not apply to the United Kingdom or any other country where such provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of express or implied warranties in certain transactions, therefore, this statement may not apply to you. This information could include technical inaccuracies or typographical errors. Changes are periodically made to the information herein; these changes will be incorporated in new editions of the publication. IBM may make improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time without notice. Any references in this information to non-ibm Web sites are provided for convenience only and do not in any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the materials for this IBM product and use of those Web sites is at your own risk. IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring any obligation to you. Appendix Page 55

58 Any performance data contained herein was determined in a controlled environment. Therefore, the results obtained in other operating environments may vary significantly. Some measurements may have been made on development-level systems and there is no guarantee that these measurements will be the same on generally available systems. Furthermore, some measurements may have been estimated through extrapolation. Actual results may vary. Users of this document should verify the applicable data for their specific environment. Information concerning non-ibm products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products and cannot confirm the accuracy of performance, compatibility or any other claims related to non-ibm products. Questions on the capabilities of non-ibm products should be addressed to the suppliers of those products. All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. This information contains examples of data and reports used in daily business operations. To illustrate them as completely as possible, the examples include the names of individuals, companies, brands, and products. All of these names are fictitious and any similarity to the names and addresses used by an actual business enterprise is entirely coincidental. All references to fictitious companies or individuals are used for illustration purposes only. COPYRIGHT LICENSE: This information contains sample application programs in source language, which illustrate programming techniques on various operating platforms. You may copy, modify, and distribute these sample programs in any form without payment to IBM, for the purposes of developing, using, marketing or distributing application programs conforming to the application programming interface for the operating platform for which the sample programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore, cannot guarantee or imply reliability, serviceability, or function of these programs. Page 56

59 Appendix A. Trademarks and copyrights The following terms are trademarks of International Business Machines Corporation in the United States, other countries, or both: IBM AIX CICS ClearCase ClearQuest Cloudscape Cube Views DB2 developerworks DRDA IMS IMS/ESA Informix Lotus Lotus Workflow MQSeries OmniFind Rational Redbooks Red Brick RequisitePro System i System z Tivoli WebSphere Workplace System p Adobe, Acrobat, Portable Document Format (PDF), and PostScript are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States, other countries, or both. Cell Broadband Engine is a trademark of Sony Computer Entertainment, Inc. in the United States, other countries, or both and is used under license therefrom. Java and all Java-based trademarks and logos are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both. See Java Guidelines Microsoft, Windows, Windows NT, and the Windows logo are registered trademarks of Microsoft Corporation in the United States, other countries, or both. Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. UNIX is a registered trademark of The Open Group in the United States and other countries. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. ITIL is a registered trademark and a registered community trademark of the Office of Government Commerce, and is registered in the U.S. Patent and Trademark Office. IT Infrastructure Library is a registered trademark of the Central Computer and Telecommunications Agency which is now part of the Office of Government Commerce. Other company, product and service names may be trademarks or service marks of others. Appendix Page 57

60 NOTES

61 NOTES

62 Copyright IBM Corporation The information contained in these materials is provided for informational purposes only, and is provided AS IS without warranty of any kind, express or implied. IBM shall not be responsible for any damages arising out of the use of, or otherwise related to, these materials. Nothing contained in these materials is intended to, nor shall have the effect of, creating any warranties or representations from IBM or its suppliers or licensors, or altering the terms and conditions of the applicable license agreement governing the use of IBM software. References in these materials to IBM products, programs, or services do not imply that they will be available in all countries in which IBM operates. This information is based on current IBM product plans and strategy, which are subject to change by IBM without notice. Product release dates and/or capabilities referenced in these materials may change at any time at IBM s sole discretion based on market opportunities or other factors, and are not intended to be a commitment to future product or feature availability in any way. IBM, the IBM logo and ibm.com are trademarks or registered trademarks of International Business Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked terms are marked on their first occurrence in this information with a trademark symbol ( or ), these symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at Copyright and trademark information at ibm.com/legal/copytrade.shtml Other company, product and service names may be trademarks or service marks of others.