Unit Versus Differential Weighting Schemes for Decision Making: A Method of Study and Some Preliminary Results.

摘要

A method for generating realistic data is described, and illustrations of how the method can be used to study the efficacy of different data analysis models are given. The method is a Monte Carlo computer simulation of a multivariate process and generates a N by M data matrix where N is the number of observations and M is the number of variables or measurements. The computer program to accomplish this is outlined. Two examples of the use of the method are given. One compares the familiar multiple regression model with simple unit weighting in a well defined prediction problem. The results indicate that multiple regression is superior to unit weighting for prediction purposes, but the differences between the two models are not great. The second example compares several ways of forming weighted and unweighted composites in a multi-attribute decision making context. Some of the conditions in which differential weighting is important in such contexts are specified. (Author)