A microarray cancer classification technique based on discrete wavelet transform for data reduction and genetic algorithm for feature selection

One of the foremost generally utilized implementation for microarray data is to characterize various kinds of diseases according to gene expression profiles. Microarray cancer data classification has been typically a challenging task for machine learning researchers. This occurs due to the complexity and the high dimensionality regarding the associated feature space despite the limited availability of training data. Therefore, reducing the dimensionality of data is an essential strategy to significantly improve the accuracy of microarray data classification. Consequently, this paper will introduce a new approach for reducing the number of features based on the discrete wavelet transform (DWT) and a new feature selection technique using a modified genetic algorithm. This suggested methodology is used to identify the minimum set of features with the most relevant impact within a large population of features. To measure the quality and performance of our proposed method, several models of wavelet decomposition levels and different types of wavelet algorithms are adopted. Also, our method is compared to other methods that were recently published using standard data sets. The experimental results demonstrate that the suggested technique outperforms the other published cancer classification techniques within the greater part of the cases that have been investigated.