Classification of microarray data involves Naïve Bayes and dimension reduction using haar wavelet

Abstract

A general problem solving for handling microarray data is classification process with added a selection process from huge attributes. In particular, the escalated of attributes dimensionality provides a challenge to microarray handling techniques, related to microarray represents the large amount of genes expression. The multi-dependency (multicollinearity) may affect the performance when determining the parameter of classification. Many ways of solving the multicollinearity problem exists, the variable selection technique has become particularly popular. This is the method which use wavelet transformation for a few carefully selected variable and the method which regress respond variable onto a few linier combinations (components) of the original attributes. Wavelet is commonly used in image processing, spectral data using wavelet transformation have proved very successful in capturing the distinction among hyperspectral data. This paper investigates a new method of transformation data using Haar wavelet for selection processes. Our extensive study compares the selection processes using Haar wavelet transformation and Genetic Algorithm considering the selection dataset that implemented to Naïve Bayes classification. In addition, the selection-classification using Haar wavelet and Naïve Bayes describes a classification cancer and non-cancer quite well related to the accuracy of confusion matrix.