Gene classification using effective random forest bootstrap technique for predicting the gene abnormalities

Abstract

Gene classification is an increasing concern in the field of medicine for identifying various diseases at earlier stages. This work aims to specifically predict the abnormalities in human chromosome-17 by means of effective random forest bootstrap classification. The homo-sapiens dataset is initially preprocessed to remove the unwanted data. The enhanced data undergoes training phase where the appropriate and relevant features are selected by wrapper and filter methods. Based on the feature priorities, decision trees are formulated using random forest technique. The statistical quantities are estimated from the samples and a bootstrap sampling is designated. The effective bootstrap technique classifies the gene abnormalities in chromosome-17. The performance metrics are evaluated and the classification accuracy value is compared with the values of existing algorithms. From the experimental results, it is proved that the proposed method is highly accurate than the conventional methods.