Performance Evaluation of Machine Learning Algorithms in Predicting Global Warming: A Comparative Study of Random Forest, K-Nearest Neighbors and Support Vector Machine

Anisa Putri (Universitas Islam Negeri Sultan Syarif Kasim Riau, Indonesia)
Refri Martiansah (Universitas Islam Negeri Sultan Syarif Kasim Riau, Indonesia)
Qhairani Frilla F. Safiesza (Universitas Islam Negeri Sultan Syarif Kasim Riau, Indonesia)
Muhammad Fahri Abduh (Kastamonu University, Türkiye)

Indonesian Journal of Applied Technology and Innovation Science
Vol. 1 No. 2 (2024): IJATIS August 2024

Global Warming is a global warming phenomenon that has a significant impact on human health and the environment. This research aims to apply Machine Learning algorithms, namely the Random Forest algorithm, K-Nearest Neighbors (K-NN), and Support Vector Machines (SVM) in predicting global warming. First, global warming data downloaded from Kaggle via dataset is used as research material. Then, a global warming prediction model is built using this algorithm and then evaluated using criteria such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean squared error (RMSE), R2, and Confusion Matrix. Finally, based on the evaluation results, research confirms that the K-NN algorithm shows the best performance, with the highest R2 value and low prediction error compared to other algorithms, such as Random Forest which shows the lowest performance. In terms of classification, K-NN achieved the highest accuracy (96.55%) and excellent performance in the confusion matrix and classification report. Overall, the findings of this study emphasize the dominance of K-NN in this context, thereby providing a strong basis for selecting models for predicting global warming phenomena.