Comparative analysis of SVM and logistic regression for classifying diagnostic microRNA signatures in colorectal cancer

Abstract

Abstract

The selection and classification of genes are critical for determining which ones are linked to a particular illness, especially cancer. As a result, it's critical to use machine learning algorithms to analyze relevant statistical data to aid biomedical researchers and end-users in the work of selection and classification. Few researches have been done on the early diagnosis of CRC using machine learning techniques to detect biomarkers, which are very important in colorectal cancer disease diagnoses. We therefore conduct a comprehensive gene selection and classification functionality using SVM and Logistic Regression algorithms on high-dimensional datasets. The results show that under the receiver operating characteristic (ROC) curve, the SVM and Logistic Regression models' discriminative capacities for classification were 83.5% and 73.2 %, respectively. This study thus reveals that the SVM algorithm outperforms the Logistic Regression algorithm in classifying data in the detection of Colorectal Cancer.            

Keywords: Algorithm, Biomarkers, Classification, Colorectal Cancer, Disease Diagnosis