The design and development of forecasting systems has been one of my favorite interests going all the way back to my first experience with NASA at the Sohio Research Facility near Cleveland where I wrote a paper on Russian satellites. My PhD from Purdue was in AI and Pattern Recognition and throughout my career in Psychology I was engaged in solving some type of forecasting problem. My interest in retail forecasting started by accident in 1976 with Steak & Ale and continues today.
Below is a brief list of some the large-scale systems I designed in the last ten years.
- : Selects the optimum marketing strategy for a set of up to 1,000 products and an unlimited number of consumers considering cost, availability, customer profile, likelihood of purchase and other factors. Developed for Experian.
- : Predicts store performance, risk and other factors for retail chains opening new locations. Developed for multiple retail chains.
- : Simulates the behavior of each camera in a global system of 100,000 cameras to reduce the impact of camera failure and also to make future forecasts. Developed for Footfall, Inc.
- : Designs the optimum strategy for a direct mail campaign for consumers in a market, region or country based on the type of mail, sequence of mailings, response rate by household type, carryover impact from cumulative marketing, cost and other factors. The system can also be used for multi-channel campaigns. Developed for Experian/Sweden.
- : A full featured market optimization and planning system based on gravity modeling logic where the consumer's potential to use store locations decays with distance or time from the location. This system has 15 optional layers to handle a variety of complex situations that occur in real world settings. Developed for Prediction Analytics, Inc.
- : A system for managing retail locations or product distribution that takes into account sales from multiple channels including retail, Internet and Direct Mail. Developed for J Jill.
- : A system that estimates the positive or negative impact of all major chains on a specific target chain based on proximity and other factors. This is used as one component in TheRetailPlanet's property scoring system. Developed for TheRetailPlanet.com.
- : A complex system that constructs a unique model, automatically, in real time given the current parameters and then produces a forecast using a hierarchical chain of robot supervisors and data aggregators. Developed for Experian.
- : A system for collecting and filtering all conversations related to health on social media and then feed these into a real time health information system. Personal invention with Lynn Cherry.
- : A system for scoring the potential of all US Retail property for all major US retail chains. Developed for Nielsen as the Retail Location Score.
- : A system to connect the online ads provided through Double Click with users (based on user profiles, times, purpose of the ads, etc.) in order to provide the optimum response. Developed for LowerMyBills.com
- : This proprietary logic partitions the world into buckets tied to latitude and longitude where each bucket represents a storage unit roughly 1k in size. Developed for DataGenesis, Inc.