Pemodelan Spasial untuk Analisa Produksi Padi Integrasi Machine Learning

Abstract

 The growing human population has sparked interest in accurate rice production prediction. This research combines Geographic Information Systems and machine learning to model rice paddy production prediction with minimal error. Using linear regression, this research analyzes rice paddy production in 13 districts/cities in South Kalimantan with spatial and historical data as features, and rice paddy production as the target variable. Experimental results show that this approach produces accurate predictions, with the Linear Regression Algorithm having a consistent level of accuracy and R² of 0.992, explaining that 99.2% of the variation in the response variable can be explained by the predictor variable, and 0.8% is influenced by other factors. While Hulu Sungai Utara district has the highest rice production and Barito Kuala district has the lowest. This study makes an important contribution in spatial modeling for rice paddy production prediction, helping agricultural experts and decision makers identify rice paddy production potential and improve the accuracy of rice paddy production prediction.