CD Skripsi
Implementasi Shapley Additive Explanation (Shap) Pada Klasifikasi Categorical Boosting (Catboost) Untuk Meningkatkan Akurasi Prediksi Temperatur Udara Di Kota Pekanbaru
Air temperature is a weather parameter that affects various aspects of life, particularly in Pekanbaru, which often experiences extreme weather conditions. The accuracy of air temperature predictions is crucial for mitigating negative impacts on agriculture, energy, and urban planning. This study aims to improve the accuracy of air temperature predictions in Pekanbaru by using a Categorical Boosting (CatBoost) method optimized by Shapley Additive Explanation (SHAP). The data used was obtained from the Automatic Weather Station at the Department of Electrical Engineering, Faculty of Engineering, Riau University in October 2023. The dependent variable in this study is outdoor temperature (Y), while the independent variables include dew point (X1), wind speed (X2),wind gust (X3), max daily gust (X4), wind direction (X5), weekly rain (X6), humidity (X7), ultra-violet radiation index (X8), solar radiation (X9) and wind chill (X10). Before optimization, the prediction accuracy was 82.83%. After applying Shapley Additive Explanation (SHAP), the accuracy increased to 86.89% with independent variables such as max daily gust (X4), weekly rain (X6) and solar radiation (X9) contributing significantly to the prediction results. The results indicate that optimization using Shapley Additive Explanation (SHAP) on Categorical Boosting (CatBoost) effectively enhances prediction accuracy and provides deeper insights into the factors influencing air temperature changes.
Keywords: Air Temperature, Categorical Boosting (CatBoost), Shapley Additive Explanation (SHAP), Prediction Accuracy.
Tidak tersedia versi lain