An Update to the Landing Craft Air Cushion (lcac) Na Vigator Selection System Prediction Algorithm

This report updates the status of the Landing Craft Air Cushion Vehicle (LCAC) Navigator Selection System prediction algorithm. The last revision took place at the end of 1997. With the receipt of new training data, the prediction algorithm was changed in June 2000 to take advantage of these new data. Some rough estimates of the new attrition rate and the rejection rate due to the selection system may be made on the basis of the new data. The attrition rate is estimated as 17.24% and the rejection rate as 36.96%. The failure rate during training appears to be about 39.13%. If the 'rejection rate of about 37% is acceptable, then the LCAC Navigator Selection System can reduce the attrition rate from around 40% to around 17%. These estimates are based on rather small numbers and therefore are subject to substantial revision as more data accrue. The prediction of a success or failure for any given candidate by the LCAC Navigator Selection System is built on the foundation of statistical decision theOlY (SDT). The selection system is trying to make a decision about a candidate whose composite score on the test battery is known but whose training outcome is unknown. There is then, by definition, some uncertainty about the training outcome for this candidate. Uncertainty and the problem of making decisions in an uncertain world is the province of probability theory and SDT. The only mathematically self-consistent way that has been found to treat uncertainty is through probability theory. From the basic axioms of probability theory, one is able to construct a generally accepted framework dealing with uncertainty called the Bayesian approach. One·ofthe core concepts within the Bayesian approach is the predictive probability density function This paper presents some numerical examples of how a Bayesian predictive density can be calculated for the LCAC Navigator Selection System.