Implementing Analytics


Pasha Roberts

Talent Analytics, Corp

Time Series Clustering to Predict Sales Representative Growth, Performance, and Gaming

In the search for quality sales representatives, it is tempting to think that there is one successful “type” of rep. In fact, the data in this study suggest that there are multiple pathways of sales achievement.

In this study we put aside simple aggregations to do it the “hard way” – by modeling low-level sales transactions. Using big data technologies, we analyzed millions of geo-located sales records for thousands of national sales reps. The analysis revealed several well-travelled pathways – some hit the ground running, some were gradual learners but eventually excelled, some gamed the system to win, many others failed.

What can we learn, infer, and predict from these patterns? Hiring, cost, operations, and predictive implications directly follow from quantifying these paths. Participants should expect insight about sales analysis, multiple pathways to achievement, and lessons learned from operating on massive un-aggregated transactions data.


As Chief Scientist and Co-founder, Pasha is responsible for architecture, development, and algorithms for Talent Analytics. He wrote the first implementation of the software over a decade ago, and today he continues to drive new features and platforms for the company.

As is often found in data science, Pasha has decades of experience/education that span computing, quantitative, artistic, and business categories.

Pasha holds a bachelors degree in Economics and Russian Studies from The College of William and Mary, and a Masters of Science degree in Financial Engineering from the MIT Sloan School of Management. His thesis at MIT prototyped the application of advanced 3D graphics to massive financial “tick” datasets.

He has founded two companies, WebLine Communications Corporation, an web-call center enterprise software company, and Lineplot Productions, a financial visualization/animation service company.

Pasha’s passion with Talent Analytics is to develop new analytics to focus business performance, and to extend the Talent Analytics model to a useful set of software platforms. He hopes to discover new information about people and the work they do, with every new project. Follow Pasha on twitter @PashaRoberts.

Back to schedule


Thomas Siems

Federal Reserve Bank of Dallas

The Wealth from Innovation: Recognizing Long Waves of Innovation and Catching the Analytics Wave

Economists have been striving to understand the forces that drive economic growth and productivity since the dawn of economics more than 200 years ago. In this session, participants will learn about the importance of innovation in generating higher levels of productivity and economic growth, how the U.S. economy has gone through four major innovation waves since its founding, and how business analytics is helping to propel the current (fifth) wave and yet, at the same time, creating new corporate and macroeconomic benefits and challenges.
While innovative ideas and entrepreneurship are at the heart of economic growth, an environment is needed where growth can be created and sustained. Great ideas by themselves will not create an economic spark. And for innovative ideas to have the greatest impact, certain other economic drivers are needed to continue to fan the flames for long-run growth.


Thomas Siems is assistant vice president and senior economist at the Federal Reserve Bank of Dallas, and teaches operations research and management science courses in the Lyle School of Engineering at Southern Methodist University where he serves as the chief engineering economist. Siems earned a BSE in industrial and operations engineering from the University of Michigan and an MS and PhD in operations research from SMU. Siems has published more than 60 articles in various academic journals, books and Federal Reserve publications and is the National Association for Business Economics’ only two-time winner of the Edmund A. Mennis Contributed Paper Award. His work has received extensive attention from leading publications, including Financial Times, Wall Street Journal, The Economist, American Banker and Investor’s Business Daily. Siems has also authored and published five children’s picture books, including The Dangerous Pet, which poetically challenges the way the next generation thinks about debt.

Back to schedule

Madan Kumar Singh


Network to Revenue intelligence (N2R-i)

The growth in revenue for the Telcos has slowed down globally. Though the data usage has increased exponentially Telcos have not been able to benefit from consumers increased consumption of data, the benefit has largely gone to the Over-the-top (OTT) service providers.

The traditional BI, predictive and optimization models have been using customer , product , customer service data etc. extensively, but now the insights from these data sources are not sufficient to tackle current revenue and ARPU issues of CSPs.

In this talk, we’ll discuss about the usage of transactional data sources, like Passive probe data, CDRs, GGSN, SGSN, performance counters etc. in combination with traditional data sources from OSS & BSS systems. The incremental insights from these new data sources, when leveraged would change the way operators have been using analytics. We will talk about how the power of 3 – Network data, statistical algorithms and real-time execution comes together to create intelligence that is helping Telcos in creating new services, managing churn, improving campaign performance and in proactive customer service.


Madan Kumar Singh has developed and implemented innovative analytics solution for Communications Industry. He serves as Lead for Industry Unique Analytics for Telecommunication and Cable Industry in Accenture’s advanced analytics practice.

Madan has more than 13 years of rich experience in consulting and analytics across industry including Telecommunications, Financial Services and Consumer goods. He has significant leadership experience in business solutioning, client management, business development, service delivery, statistical analysis, predictive algorithms and optimization models. Madan also holds patent in Predictive Node Failure Analytics and has filed a patent for innovation in developing new data services to improve revenue of Telcos. Madan has presented his work in other international conferences including Mobile World Congress, the largest conference in the world for communications Industry.

Madan holds an MBA in Marketing and Systems from one of premier B-Schools in India and a bachelor degree in Agriculture Economics.

Back to schedule


Jen Underwood


Best Practices for Developing Real-time Dashboards

In the intensively competitive, connected, big data world, speed to insight and action have become more critical than ever before. Decision makers can no longer rely on stale, static dashboards alone. In this session, we will walk through real-world, real-time dashboard and alerting use cases for operations management, finance, marketing and sports. We will share best practices, lessons learned and tips on designing and delivering real-time data, usage of streaming data, caching and storage considerations. Using REST APIs, we will step through the process of creating an example dataset, report, visualizations and pushing data directly from a variety of applications into a sample real-time dashboard.


Jen Underwood is a Microsoft Sr. Program Manager of Business Intelligence & Analytics. She works with external groups, customers, channel partners, MVPs, BI professionals and application developers. Throughout most of her 20 year career she has been researching, designing and implementing data warehousing, business intelligence and predictive analytics solutions across a variety of vendor landscapes and industries.

Jen holds a Bachelor of Business Administration degree from the University of Wisconsin, Milwaukee and a post graduate certificate in Computer Science – Data Mining from the University of California, San Diego.

Back to schedule


Haining Yu

Walt Disney Parks & Resorts

Optimizing Room Assignment at Disney Resorts 

Walt Disney World is home to four theme parks, two water parks, 26 owned-and-operated resorts, and hundreds of merchandise and dining experiences. Every year, millions of guests stay at Disney resorts. Each property receives hundreds of reservation requests per week. Assigning rooms to resort/hotel reservations is critical to maximizing guest satisfaction and operational efficiency. The volume, variety, and uniqueness of guest preferences create a number of difficult room assignment problems.

The Disney Decision Science team is partnering with SAS to design a room assignment optimization model. The complexity and scale of the mixed-integer optimization problem prevents a traditional branch-and-cut algorithm from solving the problem within the desired optimality gap in the allotted time. SAS/OR’s decomposition algorithm exploits the block-angular constraint matrix structure and dramatically decreases run time.

We describe the collaboration, explore some of the approaches used, review selected results, and discuss our plans to capture more aspects of the problem.


Haining Yu has been with Walt Disney Parks & Resorts since 2008 and is Manager Decision Science within Department of Revenue Management & Analytics. He holds a Ph.D. in Systems Engineering from Boston University.

Back to schedule