INVESTIGATION OF CAPER (COMPUTER AIDED PATTERN EVALUATION AND RECOGNITION) TECHNIQUES USING NCAT (NAVY COLLEGE APTITUDE TEST) AND BTB-7 (BASIC TEST BATTERY 7) DATA TO FORECAST SCHOOL PERFORMANCE.
Abstract
The purpose of the research was to conduct a study of the classification of Navy personnel by means of pattern recognition techniques, employing as a data base the test and criterion scores for 1962 and 1963 NROTC Navy College Aptitude Test (NCAT) records and the Navy service schools Basic Test Battery 7 (BTB-7) records. Forecasts of student grade point average and pass/fail by CAPER techniques were made and compared with the results obtained from multiple regression techniques. Essentially identical results were obtained by both methods; this finding is attributed to the absence of nonlinear relationships in the data bases. However, corroboration was made of a finding by earlier investigators that marked improvement in selection of students for service school assignment could be obtained by using all BTB scores rather than one or two. A theorem--upper bound on classification accuracy--provides a technique that was developed and applied to the BTB-7 data base. It shows that the results obtained by both linear and nonlinear techniques approach the predicted upper bound. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1967
- Accession Number
- AD0659525
Entities
People
- Donald F. Specht
- John E. Mangelsdorf
- Richard E. Klokow
Organizations
- Lockheed Martin Missiles and Space