Review on Feature Selection Methods and High Dimensionality Reduction Techniques with Microarray Datasets in Early Prediction of Lung Cancer

Abstract

Microarray dataset is one of the biological databases that contains huge quantity of gene expression data. It is predominantly used in storing maximum biological information that is related to the measurements, search criteria and real-time data required for predicting hard-hitting diseases like Lung Cancer. The Microarray dataset comprised of complex data with enormous biological features that are complicated to be analyzed and interpreted for prediction of diseases with high efficiency. The major ardent of this paper is to review the various existing microarray datasets and techniques applied in prediction of Non-Small Cell Lung Cancer (NSCLC). The paper further explores the various existing feature selection models available in microarray datasets with their tools and techniques applied for prediction. A review on High Dimensionality Reduction (HDR) in microarray datasets was proposed and the significant methods with the algorithms was studied. The assessment was conducted based on various research works conducted during the period from 2016 to 2020. The evaluation methods like accuracy, f-score and other measures to test the level of performance was presented. The research gaps pertaining to the current research scenario has been analyzed and presented with tabulated analysis. The review analysis identified that feature extraction and HDR are highly essential for efficient normalization of microarray datasets in efficient prediction of NSCLC at the earliest stage possible. The recommended suggestions ordained that the microarray dataset preprocessing and feature analysis required novel frameworks for better outcomes in the future perspectives of Lung cancer predictions.