Accession Number:

AD0735468

Title:

Some Features of the User of Linear Discriminant Analysis for Forecasting,

Descriptive Note:

Corporate Author:

FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OHIO

Report Date:

1971-10-21

Pagination or Media Count:

24.0

Abstract:

Some problems involved in improving the model for predicting the monthly quantity of precipitation based on discriminant analysis are presented. The authors discuss selection of the optimum vector predictor. An algorithum is proposed for finding the order of preferableness and the optimum number of predictors for an alternative forecast. In the case of classification into three classes, a stochastic triangle is used in investigating the form of classification regions of both identical and unidentical losses. The proposed method is used in day to day practice. Author

Subject Categories:

  • Meteorology

Distribution Statement:

APPROVED FOR PUBLIC RELEASE