Perubahan iklim global telah memicu kebutuhan akan metode peramalan suhu yang lebih akurat dan adaptif. Penelitian ini bertujuan untuk mengevaluasi dan membandingkan performa tiga model peramalan dalam memprediksi suhu rata-rata bulanan di Kota Delhi, India. Penelitian ini menggunakan pendekatan time series forecasting berbasis statistik (SARIMA dan Holt-Winters Additive) serta machine learning (Random Forest). Selain itu, dilakukan analisis korelasi menggunakan metode Pearson Correlation untuk mengukur hubungan linier antara suhu dengan variabel iklim lain seperti kelembapan, kecepatan angin, dan tekanan udara. Penelitian ini mengadopsi metodologi CRISP-DM yang mencakup tahap business understanding, data understanding, data preparation, modeling, dan evaluation. Dataset yang digunakan bersumber dari Kaggle dan mencakup data harian dari tahun 2013 hingga 2017 yang diolah menjadi data bulanan. Hasil evaluasi menunjukkan bahwa model SARIMA memiliki performa terbaik dengan nilai MSE sebesar 1.613 dan MAPE sebesar 3.73%, yang tergolong prediksi akurasi tinggi. Analisis korelasi Pearson menunjukkan bahwa kelembapan dan tekanan udara memiliki korelasi negatif terhadap suhu rata-rata, sedangkan kecepatan angin menunjukkan korelasi positif, dan seluruh hubungan tersebut signifikan secara statistik. Hasil ini mengindikasikan bahwa variasi suhu tidak berdiri sendiri, melainkan dipengaruhi oleh interaksi kompleks antar faktor iklim. Temuan ini menegaskan pentingnya mempertimbangkan variabel pendukung dalam pengembangan model peramalan suhu agar hasil prediksi menjadi lebih komprehensif. Secara keseluruhan, hasil penelitian ini berkontribusi pada pemilihan model peramalan yang optimal untuk wilayah tropis dan mendukung pengambilan keputusan berbasis data dalam mitigasi perubahan iklim dan perencanaan kota. Abstract Global climate change has triggered the need for more accurate and adaptive temperature forecasting methods. This study aims to evaluate and compare the performance of three forecasting models in predicting the monthly average temperature in Delhi City, India. This study uses a statistical time series forecasting approach (SARIMA and Holt-Winters Additive) and machine learning (Random Forest). In addition, a correlation analysis was performed using the Pearson Correlation method to measure the linear relationship between temperature and other climate variables such as humidity, wind speed, and air pressure. This research adopts the CRISP-DM methodology which includes the stages of business understanding, data understanding, data preparation, modeling, and evaluation. The dataset used is sourced from Kaggle and includes daily data from 2013 to 2017 which is processed into monthly data. The evaluation results show that the SARIMA model has the best performance with an MSE value of 1.613 and a MAPE of 3.73%, which is classified as a high accurate prediction. Pearson's correlation analysis shows that humidity and air pressure have a negative correlation with temperature, while wind speed shows a positive correlation, and all of these relationships are statistically significant. These results indicate that temperature variations do not occur independently, but are influenced by complex interactions between climate factors. These findings emphasize the importance of considering supporting variables in the development of temperature forecasting models in order to produce more comprehensive predictions. Overall, the results of this study contribute to the selection of optimal forecasting models for tropical regions and support data-driven decision-making in climate change mitigation and urban planning.
English
Perubahan iklim global telah memicu kebutuhan akan metode peramalan suhu yang lebih akurat dan adaptif. Penelitian ini bertujuan untuk mengevaluasi dan membandingkan performa tiga model peramalan dalam memprediksi suhu rata-rata bulanan di Kota Delhi, India. Penelitian ini menggunakan pendekatan time series forecasting berbasis statistik (SARIMA dan Holt-Winters Additive) serta machine learning (Random Forest). Selain itu, dilakukan analisis korelasi menggunakan metode Pearson Correlation untuk mengukur hubungan linier antara suhu dengan variabel iklim lain seperti kelembapan, kecepatan angin, dan tekanan udara. Penelitian ini mengadopsi metodologi CRISP-DM yang mencakup tahap business understanding, data understanding, data preparation, modeling, dan evaluation. Dataset yang digunakan bersumber dari Kaggle dan mencakup data harian dari tahun 2013 hingga 2017 yang diolah menjadi data bulanan. Hasil evaluasi menunjukkan bahwa model SARIMA memiliki performa terbaik dengan nilai MSE sebesar 1.613 dan MAPE sebesar 3.73%, yang tergolong prediksi akurasi tinggi. Analisis korelasi Pearson menunjukkan bahwa kelembapan dan tekanan udara memiliki korelasi negatif terhadap suhu rata-rata, sedangkan kecepatan angin menunjukkan korelasi positif, dan seluruh hubungan tersebut signifikan secara statistik. Hasil ini mengindikasikan bahwa variasi suhu tidak berdiri sendiri, melainkan dipengaruhi oleh interaksi kompleks antar faktor iklim. Temuan ini menegaskan pentingnya mempertimbangkan variabel pendukung dalam pengembangan model peramalan suhu agar hasil prediksi menjadi lebih komprehensif. Secara keseluruhan, hasil penelitian ini berkontribusi pada pemilihan model peramalan yang optimal untuk wilayah tropis dan mendukung pengambilan keputusan berbasis data dalam mitigasi perubahan iklim dan perencanaan kota. Abstract Global climate change has triggered the need for more accurate and adaptive temperature forecasting methods. This study aims to evaluate and compare the performance of three forecasting models in predicting the monthly average temperature in Delhi City, India. This study uses a statistical time series forecasting approach (SARIMA and Holt-Winters Additive) and machine learning (Random Forest). In addition, a correlation analysis was performed using the Pearson Correlation method to measure the linear relationship between temperature and other climate variables such as humidity, wind speed, and air pressure. This research adopts the CRISP-DM methodology which includes the stages of business understanding, data understanding, data preparation, modeling, and evaluation. The dataset used is sourced from Kaggle and includes daily data from 2013 to 2017 which is processed into monthly data. The evaluation results show that the SARIMA model has the best performance with an MSE value of 1.613 and a MAPE of 3.73%, which is classified as a high accurate prediction. Pearson's correlation analysis shows that humidity and air pressure have a negative correlation with temperature, while wind speed shows a positive correlation, and all of these relationships are statistically significant. These results indicate that temperature variations do not occur independently, but are influenced by complex interactions between climate factors. These findings emphasize the importance of considering supporting variables in the development of temperature forecasting models in order to produce more comprehensive predictions. Overall, the results of this study contribute to the selection of optimal forecasting models for tropical regions and support data-driven decision-making in climate change mitigation and urban planning.
Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.