Simulation of climate model is used to produce climate models used to estimate climate in the future using some software. Simulation of climate model has two probability, they are success or failure. The problem is when the simulation is fail. There are 18 variables that used to predict the simulation. Feature selection is used to reduce the dimension of variables using RFECV method. There are 11 variables that important to simulation of climate. There are 46 from 540 simulations that fail. Furthermore, SMOTE is used to handle imbalance cases. The classification method used in this paper are Logistic Regression, Naïve Bayes, Support Vector Machine (SVM), and Random Forest. The AUC value were not significantly different for the four methods when using SMOTE. However, the highest AUC was obtained by SVM method, so the simulation of climate model can be predicted by SVM method. Keywords: AUC, SMOTE, RFECV, Logistic Regression, SVM, Random Forest, Naïve Bayes
Copyrights © 2023