Analyzed results of surveys collected over four-year period containing 18,809 responses about patient satisfaction for major urban hospital and identified factors to monitor in order to improve overall patient satisfaction.
Working with AIG Inc., New York, on predicting closing of 10b-5 securities class action cases.
Working with Lockheed Martin Space Systems Co., Sunnyvale, California, on principled way of fusing multisensor information using valuation-based systems.
Analyzed performance data of agents in a Fortune 500 company’s call centers and developed algorithm to classify agents into homogenous groups. We compared differences between the groups to determine strategies to improve the performance of the poorer performing groups.
Analyzed complex business problems and developed models to forecast business and economic data, including the following examples:
Developed model using weather conditions data to forecast short-term loads that would be required for an electrical utility. This model was used by the utility in planning for resources that would be required to provide adequate power to its customers.
Developed model for a transporter using historical data to determine resources that would be required to insure proper number of vehicles and drivers were available each day to do meet the transportation requirements.
Developed Matlab® software for Raytheon Missile Systems, Tucson, Arizona, to construct and deploy expert systems in missile defense using the Dempster-Shafer theory of belief functions. The software produced, called Belief Function Machine, can be freely downloaded.
Assist clients using economic value added (or EVA) approaches to determine pricing levels for products with innovative new features, including the following examples:
Found acceptable pricing levels of client’s new ingredient after benchmarking perceived value of similar ingredients.
Worked with companies with spatial data to forecast demand and assist with retail site selection.
Analyzed historical sales force data to set sales quotas that were both motivating and cost-effective. Using past sales and quota data, the goal of this project was to identify optimal quotas for future periods. A key insight from the study is the observation that quotas were set mechanically across the entire force. Consequently, we advised the client to set quotas that were individually tailored.
Worked with broadcast television company to identify when inventory of commercial times should be sold given the pricing and the show performance uncertainties. Also, developed algorithms to schedule commercials as per client requirements, resulting in application for U.S. patent.
Worked with network-planning group of major U.S. airline to analyze network delay propagation and developed model that is computationally very fast. Also, constructed algorithms that optimize block-time allocation for given target scheduled on-time probabilities for each flight in the network. Analysis showed that this particular U.S. airline’s total network delays could be significantly reduced without increasing the block-time budget.
Study was completed in September 2012; published paper can be downloaded here.