An optimised gene selection approach using wavelet power spectrum.

Data mining is a boon to many fields like bioinformatics for processing a vast amount of data. In our previous paper, we proposed a novel feature selection method for microarray data classification using Wavelet Power Spectrum (WPS). In this paper, we present optimisation techniques to improve the quality of the features thus selected and to select 'tight genes' from various cancer microarrays. The results show that 'tight genes' thus selected were more qualitative and could be used for a wide variety of data sets. Also, 'tight genes' thus selected in this mining process could be used with any existing classification approach.