Salt Hill Statistical Consulting assists corporate clients with a wide range of issues, including assisting in marketing analytics, forecasting sales and costs, assisting with data and technology audits, compliance matters, and litigation support.
On behalf of a Fortune 100 company, we created statistical models to determine the probabilities and likely severities of accidents for different employee and accident types. Used statistical software, including the Stata program and the R Language. By focusing on the areas and types of accidents, we estimated that the company could save approximately $3 million annually in costs.
Statistical analysis and modeling on behalf of Pfizer in three separate projects over the course of several years. Validated revenue forecasting models, and wrote and implemented code in the R language to improve the forecasting process and underlying diagnostics. In another project, reviewed Health Insurance Portability and Accountability Act (HIPAA) requirements to determine if certain data sets had been properly “de-identified.” In a third project, assisted in validation of a data supplier by performing statistical analyses of possible biases in data regarding prescriptions and market share of pharmaceutical products.
In a number of projects on behalf of universities, analyzed student loan data to determine cohort default rates and other key indicators of loan performance. Some of the work was to assist in reporting to the Department of Education, while other aspects of the work was for internal control.
In separate projects on behalf of two leaders in the online business to business auction market for insurance leads, developed statistical models and analyzed data in order to optimize bidding and pricing of leads in the primary and secondary market.
On behalf of Fortune 100 insurance company, designed a statistical sample to compare rates among possible providers in different localities. Performed analysis at the patient level and at the procedure code level to help determine which provider was the most cost-effective.
On behalf of Fortune 100 company, reviewed analysis of settlement data for thousands of asbestos cases, in order to determine whether certain claims regarding company actions were valid. Performed statistical analyses to determine whether differences existed in different types of settlements. Dr. Salzberg wrote expert reports setting forth his findings.
Created probability models and performed statistical analysis in order to determine whether individuals listed in a social security database (containing millions of records) matched individuals holding insurance policies that may have required payments. Helped design software that was successful in recovering insurance dollars on behalf of the states.
In a matter regarding selling of hair products to unauthorized distributors, analyzed data on thousands of buybacks. On behalf of L’Oreal, performed statistical analysis and determined whether one company was responsible for a disproportionate amount of unauthorized sales. Dr. Salzberg wrote an expert report and testified in a deposition in the matter, which eventually settled.
On behalf of a major data provider in the pharmaceutical industry, reviewed data and developed pilot forecasting methods, which took into account the possibility that a major drug store chain might stop providing information. Worked with dataset containing millions of lines of data regarding prescriptions of individual doctors on a weekly basis across the United States.
On behalf of Xerox, analyzed text mining software and suggested algorithms to improve categorization of documents. We analyzed a model the scored documents according to whether they were relevant for a particular search. Using the scores from a proprietary process created by Xerox, we created an iterative process that re-scored documents based on a re-weighting that considered the initial score and whether a sample set of documents was, in fact, relevant. The resulting scores improved performance of the proprietary model on a new set of documents.
We developed a model to determine the profitability of television advertising by station, time of day, and day of week for an online seller of software. This model was successfully used to determine how to allocate television advertising dollars. For another client, in a different project, we analyzed and suggested improvements to a similar model for mail advertising. These improvements allowed for reduction in spending and increases in revenues.
Consulted in a litigation matter involving a global engineering and construction company and one of the world's largest energy companies, developed statistical models to estimate the probability of bolts failing along an undersea pipeline, causing a catastrophic leak of oil or gas.