Hybrid feature selection approach using bacterial foraging algorithm guided by Naive Bayes classification

Feature selection is an important technique, which applied in those areas where amount of data is too large to analyze and it becomes crucial to minimize large datasets. Bacteria Foraging Optimization (BFO) algorithm is an optimization algorithm, which get its inspiration from Escherichia Coli bacteria. This work proposed a hybrid approach of BFO algorithm and Naive Bayes classification. To evaluate the performance of new hybrid algorithm, experiment was conducted, in which five benchmark datasets were used and comparisons are done with other feature selection algorithms. Number of features and classification accuracy were major criteria for comparison. The results of experiment are evident that our algorithm outperforms the other algorithms in selecting lower number of features through comparable or better classification accuracy.