top of page

SAMPLE PROJECTS

Sales and Revenue Maximization for a Global Investment Services Company
 

Created a multi source-system, big data driven model development and deployment environment that yielded $.5 billion in annual incremental sales revenue.  The solution included the following automated features:

  • Development Platform

    • Data standardization and conversion to linear expression of dependent variable

      • Data transformation catalogs stored

    • Re-expression of missing values

    • Optimized independent variable reduction

    • Model fitting and bootstrapping

      • Predictive algorithms stored

  • Deployment Platform (Scoring)

    • Repeated source data transformation using stored catalogs

    • Repeated scoring cycles with updated data

    • Self-service tables to deliver scores for applications supporting email/direct mail (e.g. Exact Target, Unica), personal contact management (e.g. SFDC, BaseCRM), and web personalization (e.g. Blueconic)

Gas Pipeline Flow Optimization for a Global Energy Company

Built a Scheduling Optimization Tool to manage the flows and maximize returns from the transport of demethanized gas mixtures throughout a North American network of underground pipelines. Included instantiation of the following into $12 million data warehouse:

  • Models yielding forecasts of price shifts in the energy commodities market and simulating the effect on profitability of various positions assumed in response to the predicted price changes.Included the measurement and simulation of risk associated with trading natural gas and energy derivatives under varying market conditions using methods such as value-at-risk and Black-Shoales.

  • Models optimizing routes, batch sequencing, batch sizes, flow rates, scheduling and customer demand satisfaction for a 10,000 mile network of natural gas liquid pipelines using mixed integer linear programming, network optimization, heuristic and non-linear programming and simulation techniques.

  • Models predicting the relative distribution of multiple components (e.g. ethane, propane, butane, etc.) in mixed batches of natural gas liquids based on historical gas chromatograph readings in association with inter-pipeline pressure, temperature and density. Employed linear and non-linear statistical methods and multi-layer backpropagation neural networks.

  • Models forecasting the demand for propane in various markets based on weather, crop yields and historical demand.Employed linear and non-linear statistical methods and multi-layer backpropagation neural networks.

Global Adverse Event Compliance Reporting Simulation and Control for a Major Pharmaceutical Company

 

Collaboratively designed, constructed, and deployed a highly sophisticated stochastic model that simulated global adverse event reporting processes for a major pharmaceutical company.  The model is used to test the economic and operational viability of various strategies including outsourcing, staffing patterns, and numerous processing protocols.  The model insured compliance with myriad reporting requirements around the globe thereby minimizing costs by insuring optimal dispersion of resources and avoidance of non-compliance penalties levied by international authorities.

Software Product Bundling and Online Promotional Optimization for the Leading Web-Based ASP Managing Transactions Between Car Purchasers and Dealerships
 

Managed data warehouse design and data mining initiatives for a large web-based ASP that provides software and services to manage the transaction between automotive customers and lending institutions

  • Conducted market basket analyses for bundles of software products. Included the development of complex SAS programs to measure the time-neutral effect of ancillary product adoption on loan application volume, which was the primary source of income for the company.  This provided guidance for product bundling and promotional programs.

  • Created loan application demand forecast models.  Used time series models including ARIMA and dynamic regression with exogenous factors.

  • Analyzed web logs to ascertain traffic patterns consistent with desired behavior, e.g. loan application submission, and modified online navigation cues to encourage optimum routing to landing pages.

bottom of page