<![CDATA[Penelitian ini menyajikan analisis komparatif model pembelajaran mesin untuk prediksi banjir menggunakan data historis curah hujan. Tiga model, yaitu K-Nearest Neighbors (KNN), Naive Bayes, dan Random Forest, dievaluasi berdasarkan metrik kinerja mereka. Evaluasi mencakup akurasi, presisi, recall, skor F1, dan ROC AUC. Hasilnya menunjukkan bahwa model Random Forest mengungguli yang lain, mencapai skor sempurna dalam semua metrik. Namun, KNN dan Naive Bayes juga menunjukkan kinerja yang kompetitif, meskipun dengan beberapa trade-off antara presisi dan recall. Temuan ini memberikan wawasan berharga tentang efektivitas berbagai pendekatan pembelajaran mesin untuk prediksi banjir, yang berkontribusi pada pengembangan sistem prediksi banjir yang lebih andal. This study presents a comparative analysis of machine learning models for flood prediction using historical rainfall data. Three models, namely K-Nearest Neighbors (KNN), Naive Bayes, and Random Forest, are evaluated based on their performance metrics, including accuracy, precision, recall, F1 score, and ROC AUC. The evaluation results show that the Random Forest model consistently outperforms KNN and Naive Bayes. Random Forest achieves a perfect score (100%) on all measured indicators. Meanwhile, KNN and Naive Bayes also demonstrate competitive performance, albeit with some trade-offs between precision and recall. Specifically, for accuracy, precision, recall, F1 score, and ROC AUC, the Random Forest model scores 100%, whereas KNN and Naive Bayes are in the range of 90-95%. Nevertheless, KNN and Naive Bayes still show competitive performance and are worth considering as alternative flood prediction models. These findings provide valuable insights into the effectiveness of various machine learning approaches for flood prediction. The Random Forest model proves to be the superior approach, yet KNN and Naive Bayes also show significant potential. The results of this study contribute to the development of more reliable and accurate flood prediction systems, with important implications for disaster management and flood risk reduction..]]>
Copyrights © 2023
