RI 9339 - Genetic Algorithm Applied to Least Squares Curve Fitting

The U.S. Bureau of Mines is currently investigating the use of genetic algorithms (GA's) for solving optimization problems. This computer search technique, based on the mechanics of natural genetics, was used to perform simple least squares curve fitting. Three examples are presented in which a GA manipulates binary coded strings to produce near-optimal solutions to least squares curve-fitting problems after having viewed only a small portion of the search space. The examples include fitting data to the equation of a line, the equation of a parabola, and the Ree-Eyring equation.