A Framework for Selecting Features using Various Soft Computing Algorithms

“Particle Swarm Optimization” (PSO), “Artificial Bee Colony algorithm” (ABC) are the typical example of evolutionary algorithm. In this paper pre-processing was done initially to clean the data then proposed “Grouped Artificial Bee Colony Optimization” (G-ABC) algorithm for optimizing the feature selection procedure. Proposed feature selection model includes combining employee bees to generate more efficient rule generation with less time of execution. Furthermore, experimental results were analysed and verified. The experiment compares basic PSO, basic ABC, PSO- ABC and G-ABC (proposed model) algorithms on baseball data set. The results indicate that G-ABC with Neural Network algorithm has good performance in term of selecting features(selected 91 features out of 100 records), time of execution(executedin 3 sec) and 98.2% accuracy.