(2010,  Revenue Management Systems, Inc )

Assessed different statistical models for their accuracy in predicting demand for airline tickets of different prices based on the number of days before the flight.

Statistical Models Used for Forecasting: Exponential Smoothing, ARIMA at Wikipedia

Assessment Measures for the Models: MAPE, SMAPE, MASE (Make these links Wikipedia)

(2005, T-Mobile)

Built a set of SAS macros to input data from many sources, clean and combine the data, generate data reports, and generate forecasts for call volume by date, time of day, and holidays. Assessed the accuracy of different statistical models for these forecasts.

Statistical Models Used: Exponential Smoothing, ARIMA, Multivariate ARIMA

Assessment Measures for the Models: MAPE, SMAPE

(2011, Oregon  Department of Consumer and Business Services)

Determined the effectiveness of OSHA inspections on reducing workers’ compensation  claims.

Statistical Models Used: Poisson Regression, Negative Binomial Regression,

Zero-Inflated Poisson Regression, Zero-Inflated Negative Binomial Regression

(2008, Visa)

Performed several statistical analyses to determine whether or not a highly-targeted direct marketing program to high-spending customers significantly increased the average spending per account and the average number of transactions per account.

Statistical Models Used: Fisher’s exact test, t-test, and Pearson’s Chi-squared test.

(2005, T-Mobile)

Created a data repository for call center data to analyze the impact of new initiatives on call center demand. These new initiatives included text messaging for bill paying and speech recognition software to answer caller questions without involving a customer service representative.

Statistical Models Used: Multivariate ARIMA

(2010, Looking Glass Analytics)

Used advanced statistical techniques (Empirical Bayesian methods) to estimate the incidence rates of various diseases in sparsely-populated regions of Washington State.

Statistical Models Used: Empirical Bayesian methods 

(2009, Looking Glass Analytics)

Derived formulas for computing age-adjusted incidence rates for various diseases and other conditions. Wrote code for implementing those formulas in a SAS macro.

Statistical Models Used: Binomial, Poisson and Chi-squared distributions.

(2008, QL2 Software)

Estimated the aftermarket (reselling/scalping) sales of tickets for various venue events, based on online prices listed on various websites. Developed a data-gathering algorithm to aggregate data from different sources and assess their quality.

(2010, Fabric Worldwide)

Produced graphs of click-through rates on various ads over time and used those graphs, along with a specialized algorithm, to create working hypotheses of where users were coming from for different times of day and days of the week--and to try to predict where they might come from in the future.

Statistical Models Used: Logistic regression

(2010, Intel)

Developed an algorithm for determining what items were bought together using an industry-standard method for computing the correlations between pairs of items.

(2007, Washington Mutual)

Used SAS to create an automated process for transferring and consolidating interest rate data from multiple Excel spreadsheets in different formats into a single internal database. 

(2007, Washington Mutual)

Used SAS to create an automated process for inputting data from several different files and data formats into multiple customized Excel spreadsheets that produced graphical summaries of monthly loan portfolio data.