Application of microcomputer-based robust regression methods to nonlinear data analysis

Abstract

Two nonlinear robust regression methods are developed, characterized, and compared for the example case of exponential decay. Extension to other nonlinear models is obvious. The methods are computationally simple and easily programmed. The resulting programs exhibit reasonable computation times on microcomputers and are thus an easily implemented analytical tool. Programming examples of the robust regression and other methods are given for GAUSS, a particularly convenient language for algorithms that make extensive use of matrix manipulations