Chemical compounds can be distinguished into active compounds or commonly called bioactive compounds and inactive compounds or commonly called passive compounds. At this time there are still many active compunds that the pharmacological role does not known yet, so the system being made for classify the functions of active compounds that expected to support chemists research in the laboratory. To simplify the process of making the system, the representation of molecular structure must be easily processed by a computer so that the SMILES notation will be used, the SMILES notation describes chemical formula in a row notation. This system is using the SVM (Support Vector Machine) method because the SVM method has high generalization capabilities without requiring additional datasets. In this research uses as many as 15 features and objects as many as 3 classes of active compound functions, including metabolism, infection, and anti-inflammation. The best test result is 83.33% when using the Gaussian kernel RBF, using a lambda value (λ) of 5, the complexity value is 0.1, the sigma value (σ) is 0.5, and with the number of iterations is 5.
Copyrights © 2019