Optimizing lineage information in genetic algorithms for producing superior models

Abstract

A lot of research in the area of genetic algorithms (GA) is applied, but little research examines the impact of lineage information in optimizing a GA. Normally, researchers consider primarily elitism, an approach which carries only a very small fixed subset of the population to the next generation, as a lineage strategy. This paper investigates several different lineage percentages (what percent of the population to carry forward) to determine an ideal percentage or range from improving the accuracy of a GA. Several experiments are performed, and all results are statistically validated.