Microcomputer implementation of robust regression techniques

Abstract

Systems Engineering requires that relationships between variables under control (independent variables) and variables which are to be controlled (dependent variables) be quantified. The most common method for characterizing this relationship is to use least squares regression. This method is inadequate in cases where some of the observed relationships between variables is contaminated with aberrant data. This project provides two microcomputer-based tools that will aid the systems engineer in identifying such instances of aberrant observations. Although mainframe implementations of these tools exist, microcomputer implementations of these tools are needed to accommodate many engineers which find microcomputers more accessible than mainframe computers.