Results and discussion . Optimal machine learning algorithm selection . To rigorously select the suitable machine learning algorithm for predicting the anti-commensal effect of drugs, six classical and commonly used machine learning algorithms (SVM, k -NN, RF, NB, GBM, and XGB) were employed and compared directly. Twenty-four sets of different molecular features were generated for all 391 anti-commensal compounds and 790 commensal compounds in the dataset. The performance of each mode...