In the Automobile Industry, one of the major challenges faced by the dealers is to understand the priorities of the customers before purchasing a particular car. In the current Sci-Fi age this task is even more challenging as selling a car no longer depends on the inventory, nor price and reputation. In this Capstone project, we are statistically mining consumer survey data and web warehouse data collected in last 3 years by Cobalt (a digital marketing company owned by ADP). The purpose is to better understand the needs of automotive shoppers at different stages in the car shopping process. In our analysis, we connected multiple data sources and built a statistical model that enables the dealers to categorize website visitors into different categories (like new car/used car) of shoppers based on their browsing behavior. The model has been rigorously analyzed and tested and it has proved extremely important and helpful in improving targeted marketing and advertising for Cobalt.

The Road Map Project is a collective-impact initiative whose mission is to double the rates of post-secondary degree attainment by 2020 and to close the achievement gaps present in South King County by aligning the efforts of government and communities, of which data plays a critical role. In 2013 the RMP sought to improve its existing folder-based data management by building an analytical data system. The work consists of three phases: building an operational data store, building the data warehouse, and developing applications for reporting and exploration. Organizationally, the system will allow for more automated reporting and data management, freeing up time to engage with stakeholders and develop exploratory analyses and predictive models of student success. More broadly, the impact of this work will empower parents, communities and districts to better explore and measure educational outcomes thereby building account ability and more rapidly aligning efforts at achieving results.

ADP Cobalt group, a leading provider of digital marketing solutions for the automotive industry, has spent years collecting customer survey feedback on their dealer websites. The company now wants to leverage this valuable data in order to gain meaningful insights that will help guide decisions to optimize content and website functionalities for the end user. Our team will execute a quantitative research project using advanced natural language processing tools to help systematically categorize the unique open text responses and statistical learning methods to cluster customer segmentations, these findings will potentially be used by business stakeholders and the research team to aid in product optimization decisions.

Millions of people are suffering from cancer and tumors. We need to make a difference! With dedication to ensure patients can receive the most effective treatment possible, the Presage online monitoring system are developed to improve the cancer drug development process and gap current technology limitations. The web application provides online data monitoring by enabling scientists to cross check data and ensure both accuracy and precision of predicted results by switching different classification algorithms. Also, they are able to interact with charts and graphics demonstrating historical data comparison and distribution. Even without accessing the app, they will receive highly customized digest reports containing details about outlier images. We make information intuitive and innovative.

Expeditors® is a global logistics company with over 13,000 employees in a worldwide network of over 250 locations across six continents. Much of Expeditors business is based on the usage of event code, which enable Expeditors and customers to track freight/shipments and related documents. Our team worked closely with Expeditors Global Data Strategy team to organize the confusion around event codes and identify opportunities to improve the business. Our capstone team conducted a sequence analysis on different shipment levels. By answering the question ‘Where is my freight,’ we created rules to identify the shipment errors from both business intelligence and logical perspectives. In order to organize the event code metadata, we categorized all the event code by logical grouping and product/service perspectives. Then we delved further into the integrity of the core events, and conducted data quality analysis. The whole analysis process enables Expeditors to make better decisions as a company on how they want to manage their processes and to provide better service to customers.

Today’s Extract, Load and Transform (ETL) tools focus on graphical design and representation of the process, leaving the required mapping and transformation identification to be done manually. This is extremely difficult when faced with multiple, varied data sources. EZ-ETL uses ontology-driven mechanisms to translate data from multiple sources into concepts specific to the customer knowledge domain. OKDoctor uses the EZ-ETL frame work to evaluate medical research in terms of validity and relevance, providing the Validity Potential Rating as a novel way to support practitioners of Evidence Based Medicine. These evaluation techniques are equally applicable to other industry verticals such as finance, national security & intelligence, and legal.

Alaska Airlines has a large pool of 3300 flight attendants. To improve the work efficiency and lower down the labor cost, there is an immense need for supervisors to measure the performance of flight attendants on the basis of Key Performance Indicators (KPIs). It is also very important for every flight attendant to be able to track his performance under different performance categories as well as view the average peer score for comparison. There should be an easy, efficient, and less time-consuming way to do this task as the major problem is that the data is scattered at different places. Our Project: It involves researching and refining the data available from different sources. It also involves designing a prototype of a user friendly flight attendant performance measurement dashboard which will be able to fetch relevant data from different places and show the processed information (performance scores) at a single platform.

With approximately 9,000 public library systems across the United States, the Seattle Public Library (SPL) faces the challenge of identifying a network of similar library systems for collaboration.  The current comparison and analytics are somewhat basic and only take into consideration traditional library measures. These measures do not capture the complexity of the diverse communities in which the libraries are rooted. The goal of this project is to build on these measures and incorporate demographic data to enable a holistic comparison. Additionally, rigorous statistical analysis methods will be applied to the data in order to improve the accuracy of results. By identifying and collaborating with similar libraries, SPL will be more effective at serving its community.

The IAL Database is a storage and retrieval system for identifying the leadership (head coaches, university presidents, etc.) and institutional movement patterns of Division I Football Bowl Subdivision (FBS) programs. Using this application, IAL researchers will be able to evaluate the average term and turnover of the leadership of FBS programs, providing UW and other institutions the opportunity to make informed decisions on future contracts. This system provides a lens for institutions, as well as the public, to begin to understand what has grown into a multi-billion dollar industry. Our team took the previously disjointed processes that included paper based records and spreadsheets, and centralized everything into one database with an intuitive web-interface. This web-application now allows the IAL staff to easily manipulate data and generate custom visualizations and reports to further their understanding of the phenomena known as the coaching carousel.