A combined analytical and empirical method to define human performance measures is required for effective automated training. In theory, performance measures can be weighted and combined in a single first order equation upon which an automated training system can track a student's comparative skill level. A form of multivariate discriminant analysis was the statistical technique incorporated into the measurement selection and weighting scheme. (Vreuls and Wooldridge 1977). As with regression analysis, this discriminant technique was subject to debilitating problems such as overfit and undesirable inaccuracies in the coefficients. The author developed and implemented an adjustment to the discriminant function analogous to the ridge regression analysis suggested by Hoerl and Kemmard (1970a). To further demonstrate the effectiveness of ridge discriminant analysis and to determine useful performance criteria the author developed a Monte Carlo Simulation which provided a relative comparison of various performance measurement models. Two sets of human performance data were used to demonstrate this new statistical tool.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:rtd-1528 |
| Date | 01 January 1980 |
| Creators | Wooldridge, Lee |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Retrospective Theses and Dissertations |
| Rights | Public Domain |
Page generated in 0.0019 seconds