Non-Linear Regression Analysis: A General Program for Data Modeling Using Personal Microcomputers.

摘要

This report documents a general non-linear regression program for fitting data to non-linear models. The program is based on an algorithm which uses a least squares criterion to calculate successive improvements to an initial set of parameter estimates. The program is written in the BASIC language common to most microcomputers, because it is easy to use and to transport between machines from different manufacturers. The majority of inexpensive microcomputers do not offer matrix operations as part of their BASIC interpreter. The program presented here, therefore, supplies subroutines in BASIC for the zeroing, transposing and inverting of the required matrices, to make it compatible with most microcomputers available today. The report gives examples and program output based on a demonstration data set involving antigen-antibody complexation in solution. Two derivations of function subroutines are given to assist the user in developing his own functions. A complete listing of the necessary programs is given along with a section on program cautious.