Modeling & ETL:

The Heart of the Matter

My earlier career had a sharp focus on building predictive models and gathering the data underlying them. This included working for a mortgage insurer, a city government, and two utilities (telecommunications and electric).

​

At Mortgage Guaranty Insurance Corp. as Risk Management Manager:

• Modeled mortgage loan portfolio risk for default, foreclosure and prepayment on internal and client bank loan portfolios;

• Developed logistic and neural network qualitative outcome models using SAS® and Neuralware® software;

• Sourced input data from internal databases and external electronic sources (e.g., U.S. Treasury and BLS, WEFA, DRI, etc.)

• Wrote ETL processes to prep data for modeling and drop scores into database.

• Deployed loan default models into Freddie Mac’s loan evaluation software.

• Managed the development of:

  • Multi-bank loan database from which to develop client bank risk models.

  • Lender Landscape®, an interactive loan portfolio risk reporting system with SAS® back end.

• Unix®, Windows® and IBM® mainframe TSO® environments.

At City of Milwaukee Budget Office as Urban Economist:

• Responsible for state shared revenue budget and formula analysis, capital and debt budgets.

• Integrated TSP® and SAS® forecasting into budget and shared revenue analysis processes.

• ETL involved transcribing input data from paper historical budget documents.

• Authored city’s public pricing user fee policy used as rationale for moving sewer and other funding from property tax to user fees.

• Windows® and IBM® mainframe CMS® environments.

At Ameritech, Inc. (Regional Bell Operating Company):

• As Campaign Analytics Manager:

  • Designed, modeled, executed and evaluated quarterly residential and business telecommunications service marketing, retention and win-back campaigns;

  • Campaign revenues ranged from $20 to $85 million;

  • Executed on large datasets for 15 million customers;

  • Worked heavily with SAS® logistic, neural network and tree models;

  • Advised marketing team on appropriateness of offers, cells, test cells, etc.;

  • More streamlined ETL processes, with internal data extracted from databases and electronic transfer of external data;

  • Windows®, Unix® and IBM® mainframe TSO® environments.

• As Demand Analytics Economist:

  • Estimated demand, revenue, cost and class of service choice models for corporate planning and state and federal regulatory filings.

  • Composed testimony and performed rate impact analysis for models used in regulatory filings (state and federal);

  • Highly manual ETL processes form historical reports;

  • Team accurately forecasted local access minutes demand in the face of historic drops in long distance prices;

  • Team introduced the use of class of service choice models using logistic models;

  • Models included pooled cross-sectional time series using lagged dependent variables or polynomial distributed lags, and logistic (single and nested);

  • Used SAS® on IBM® mainframe TSO® environment and SHAZAM® on Windows®.

At We Energies as paid Graduate Summer Intern:

• Developed SAS® Box-Jenkins demand models and related graphics for farm sector electricity demand (IBM® mainframe TSO® environment).

• Extremely manual ETL processes, involving entering data from historical reports, accounting journals, etc.;

• Developed Marginal cost calculations for generating, transmitting and distributing electricity (in IBM® Lotus Symphony® spreadsheet), using what was called the CGS method (see Cicchetti, Charles J., W. Gillen and P. Smolensky. The Marginal Cost and Pricing of Electricity: An Applied Approach. Ballinger Publishing Company, 1977.)