Open AccessPERBANDINGAN BEBERAPA METODE KLASIFIKASI DALAM MEMPREDIKSI INTERAKSI FARMAKODINAMIK
Author(s) -
Hasnita Hasnita,
Farit Mochamad Afendi,
Anwar Fitrianto
Publication year - 2020
Publication title -
indonesian journal of statistics and applications
Language(s) - English
Resource type - Journals
ISSN - 2599-0802
DOI - 10.29244/ijsa.v4i1.328
Subject(s) - support vector machine , random forest , mathematics , resampling , similarity (geometry) , artificial intelligence , pattern recognition (psychology) , statistics , computer science , image (mathematics)
One mechanism for Drug-Drug Interaction (DDI) is pharmacodynamic (PD) interactions. They are interactions by which the effects of a drug are changed by other drugs at the site of receptor. The interactions can be predicted based on Side Effects Similarity (SES), Chemical Similarity (CS) and Target Protein Connectedness (TPC). This study aims to find the best classification technique by first applying the scaling process, variable interaction, discretization and resampling technique. We used Random Forest, Support Vector Machines (SVM) and Binary Logistic Regression for the classification. Out the three classification methods, we found the SVM classification method produces the highest Area Under Cover (AUC) value compared to the other, which is 67.91%.