Tyco Data Set Integration Project Design Team Naif Al-Mulhim, Samuel Bar, Sunny Gupta, Rebecca Payne Design Advisor Prof. Abe Zeid Project Sponsor Edward Jones Abstract Many Tyco products rely on dynamic data sources to deliver value to customers. This information comes from access control, video recording, point of sale, inventory, and fire protection systems. At present this information is stored in three separate databases; the CCure 9000 for security, NaviStor for point sale, and TrueVue for inventory. Though all these databases use locations to classify events, they define location in different ways. This prevents the separate sets of data from being integrated and Tyco s customers from reaping the benefits of the insights resulting from the collective information. In order to integrate these datasets, developing a robust universal definition of location (UDL) was essential. This required the team to analyze the schema of the three databases and determine which type of UDL was applicable. Using this, a system was developed which presents the combined datasets in one layout as well several sample applications of the newly available information. Key Fire Alarm Camera Computer (on) POS Terminal Exit
The Need for Project Tyco s partners are constantly looking for a higher level of quantity and quality of data about their consumers. Integrating their existing products with respect to location will satisfy this need. (36.424, 63.721) At present, Tyco s Retail Solutions has three database products which collect and store diverse types of data in a wide variety of ways. The common thread throughout them is that they look at location related events, such as a transaction. As with any system, the information available is generated by the quantity and quality of data available. If Tyco s customers could combine the data from their various systems, they would be able to conduct a broader range and depth of analytics and allow them to gather more knowledge about their consumers. The large retail organizations that partner with Tyco strive to know more about their consumers in order to drive higher sales and decrease costs. If point-of-sale, inventory, and access control event data was combined, new analytics could be performed which generate value. (44.465, 63.721) When designed properly, location based analytics could be done at both a micro and a macro level. One such example, the relationship between item location and sales, could only be derived by looking at the connection between the point-of-sale and inventory databases. On the micro level, an analyst could examine sales of similar items throughout the store, choosing the optimal location for each. From a macro level, heat maps representing inventory turnover could be used to identify the relationship between consumer sub-groups and inventory turnover. These types of analyses will allow Tyco s partners to increase sales, thus providing value. The Capstone Project Objectives and Requirements Project success will be judged on the ability of the team to design a process and system which integrates Tyco s three separate databases in a way to drive value to Tyco customers. Design Objectives The objective of this project is to integrate the three existing Tyco databases and visualize it in a way which allows analysts to derive new information utilizing the three databases. Due to the complex and confidential nature of the data contained in the products, this must be done without actually combining the databases, knowing the specific end-users, or accessing the actual data itself. These constraints make force the project to deliver a broad solution which can work in many environments and scenarios.
Design Requirements For the project objectives to be achieved, four criteria must be met: all the data sets are integrated properly; the data can be retrieved and visualized in one platform the platform must allow for open ended analytics to be conducted; and the project must be implementable on a large scale. Due to the subjective nature of these criteria, success can only be quantitatively assessed by using pre-defined metrics Design Concepts Considered Four different design concepts were considered: In order to achieve the stated project goals, four design concept candidates were explored: Data Warehouse Loading, Universal Location Hierarchy, Database Integration Key, and Data Independent Locations. Each was evaluated based on their ability to accurately represent the data and efficiently integrate it. In addition, visualization Data Warehouse Loading and scale ability options for each were explored. Data Warehouse Loading, the process of importing multiple data sets into a new database, was deemed impossible within the constraints of the project. Due to the volume of entries for each database (~5,000,000,000 bytes Universal Location Hierarchy per day per customer) combining them would be financially prohibitive and logistically impossible. Universal Location Hierarchy, which uses a location pyramid to classify each item as part of several sublocations, was considered as a possible solution. But upon analysis of the databases, it was seen that the lack of a Database Integration Key hierarchy in two of the databases eliminated this solution A B C D Database Integration Key, the development of a cross 1 2 3 reference table which instructs the interface how to interpret locations from two different databases, was deemed impractical. This is because its scalability would require constant maintenance and would limit the types of Data Independent Locations analyses that could be conducted. Data Independent Locations, which assigns a location to each object outside the context of the already existing data, was seen as the only viable solution. Though time intensive to implement initially, it allows for full integration, open ended analytics, and full scalability.
Recommended Design Concept With Data Independent Using the Data Independent Locations design concept, the team Locations as the design concept, developed a solution which meets the project objectives and falls the team narrowed down which within the scope of the project. This solution, Geographic Information software packages and devices System (GIS) coordinates, has several advantages and has been fully would meet the project developed. In practice, each location based item in a database (fire specifications. From the wide alarms, cash register, inventory shelf) would have GIS coordinates range available, the Ashtech associated with it. In addition, any departments, stores, or regions Mobile Mapper 6 was selected to would be defined as a set of points. Using these coordinates as an collect the data while AutoCAD, overarching primary key through which each object can be identified, a ArcGIS, and ArcPAD were map of the system, whether it be an individual department or selected to visualize the data. geographical region, will be able to show the locations of each object. Utilizing this map, analyses can be conducted to connect events in a database with locations on a map. To collect the data points, the team looked at currently available commercial GIS systems. Taking into consideration accuracy requirements, software capabilities, and cost, the teams choose the Ashtech MobileMapper 6. This device has the ability to gather data at sub-meter accuracy levels (with post processing), using Microsoft Windows Mobile, and came close to the budget set by the sponsor. Unfortunately, it was discovered that GIS coordinates cannot be accurately collected from the interior of a building, so only exterior coordinates were collected. Several different products and systems were explored which combined could produce the desired visualization results. Initially, the team considered the Google products of SketchUp and Earth due to their ease of use and cloud based performance. These were eventually disregarded because they do not easily integrate with Microsoft Access databases and lack the ability to conduct basic analyses without extensive coding. The second set of products, AutoCAD and ArcGIS, are considered the industry standard for layout and mapping applications, and were thus optimized for our projects needs. The features they contain allow for a wide array for analyses to be conducted. In addition, ArcGIS has a mobile mapper companion software, ArcPad, which streamlined the data collection process.
Financial Issues Given the need for a high accuracy GIS device, the projects sponsor provided funds for one to be acquired. In order for the project to be completed, only one purchase had to be made: the GIS device. Due to the high accuracy requirement, only a few devices met specifications and even fewer once cost was taken into consideration. As this was outside the level of funding available from Northeastern University, the project sponsor volunteered to purchase the device. For a full scale implementation though, the GIS device would not be the highest cost. As this system uses much software which was available at the university, if the sponsor were to implement it they would have to purchase licensees and server space for all the products involved. In addition, any full scale implementation would need several sets of devices and software to speed the product delivery, further increasing the cost. Recommended Improvements Future iterations of this project As the team was composed of Industrial Engineers, they system would include a more diverse set developed works only for a small set of possible applications of possible analyses and server (analyses). In order for it to move fully from concept to product, much side deployment. coding would have to be done to allow the system to handle the large quantity of data and variety of ways in which it is analyzed. In addition, the system would have to be moved to a server based system, allowing the data and analyzed to be accessed from anywhere there is an Internet connection.