Further Research into a Non-Parametric Statistical Screening System.
Abstract
A new statistical technique is introduced for screening a population on the basis of their observed characteristics. The technique treats nominal independent variables with a binary dependent variable. Different objective functions are specified for constructing different decision rules. A recommended decision rule results from achieving a proportionate reduction in error (PRE), by using information in the independent variables rather than just the dependent variables. Current approaches to screening are compared using five desirable properties postulated for decision rules. A Monte Carlo simulation approach is used to construct decision rules using Boolean operators. Finally, a General Sequential Algorithm is presented. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 14, 1979
- Accession Number
- ADA081803