Feature selection methods in microarray gene expression data: a systematic mapping study

Feature selection (FS) is an important area of research in medicine and genetics. Cancer classification based on the microarray gene expression data is a challenge in this area due to its high-dimensional features and small sample size. This can negatively impact the performance of data mining and machine learning algorithms. FS is a key issue in reducing the size of the microarray, which is done to obtain useful information and eliminate redundant features. With the absence of a thorough investigation of the field, it is almost impossible for researchers to get an idea of how their work relates to existing studies and how it contributes to the research community. This paper provides a systematic mapping study to analyze and synthesize the studies conducted on the FS techniques in microarrays. To this end, 108 related articles published between 2000 and February 2022 were selected and reviewed based on five criteria: year and region, FS method adopted, dataset type, source of release, and type of evaluation software. Our main goal is to provide a fair idea to future researchers about the current situation of the field and future directions. The results of the study showed that classification is the most important task in FS. In a history-based evaluation, evolutionary methods were found to have the widest application to FS.