Statistical Software for Bootstrapping Regression Models

Period of Performance: 09/01/1998 - 03/31/1999

Unknown

Phase 1 SBIR

Recipient Firm

Statis
Sebastopol, CA 95472
Principal Investigator

Abstract

The advent of inexpensive, powerful desktop computing has the potential to revolutionize the practice of statistical data analysis. Idealized models and assumptions can now be replaced with more realistic modeling, and in many cases, by virtually model-free analysis. The ensuing rewards are best highlighted by the recent enormous success of bootstrap methodology, which has conquered many difficult problems long thought to be insurmountable. The long-term goals of this project are to incorporate contemporary, innovative statistical methodologies into an easy-to-use comprehensive statistical software package capable of being used effectively by all levels of practitioners of data analysis. The resulting software will be made available as a stand-alone package and be distributed via the Internet. In Phase I, a prototype program for implementing bootstrap methodology in regression analysis will be developed. Regression models are the work-horse of applied statistics and biostatistics and are used extensively to great profit in a wide range of contexts. Their structure allows relatively simple analyses of complicated situations where interest lies in sorting out the effects of many possible explanatory variables on a response variable. Their wide appeal, applicability, and interpretability make thee methods relevant and useful within most, if not all, of the Institutes within NIH. PROPOSED COMMERCIAL APPLICATION The commercial potential for software for computer-intensive statistical methods is enormous. These methods provide an automatic approach to problems of inference and can successfully handle circumstances where standard approaches are deemed inappropriate. Data analysts of all levels and subject-matter needing to take advantage of these new procedures are potential customers of such a statistical software product. The commercial potential is strengthened by making the software widely and easily available via the Internet.