Implementation of Naïve Bayes Algorithm to Predict Food Crop Production Results in Garut Regency

The ups and downs of food crop production each year are caused by changes in the area of land planted each year. These changes are influenced by several factors, including crop rotation, government policies, changes in agricultural practices, environmental factors such as climate, and economic pressures. In an effort to improve the efficiency and productivity of food crop production in Garut Regency, the use of technology and data analysis methods is becoming increasingly important. This research aims to predict food crop production in Garut Regency with Naïve Bayes algorithm and evaluate influential factors. This modeling is analyzed using Feature Forward selection and SMOTE techniques to determine the most influential attributes and overcome class imbalance. The method used is Cross-Industry Standard Process For Data Mining (CRISP-DM). Where the use of SMOTE successfully handles unbalanced classes, and the application of Feature selection results in the 5 most influential factors, namely crop type, added planting, realized harvest area, realized production and production. The results showed that the Naive Bayes model with Cross validation and Xgboost resulted in an Accuracy value of 82.54%, Recall value of 81.67%, Precision value of 83.34%. And the AUC value is 0.904% with the Good Classification category.