Air Quality Classification Using Naive Bayes Algorithm With SMOTE Technique Based on ISPU Data

Air pollution in DKI Jakarta is an important issue and has a negative impact on public health. This study applies the naive Bayes algorithm to classify air quality, Utilizing the SMOTE technique effectively addresses the issue of data imbalance. The data analyzed came from air pollution index data from 2022 to 2024, taken from five air monitoring stations in Jakarta. The analysis process was carried out following the CRISP-DM stages, starting from understanding the problem to evaluating the model. The results showed that SMOTE succeeded in increasing prediction accuracy in fewer classes. Without SMOTE, the model accuracy reached 90% but appeared biased towards fewer classes, with a recall value of only 0.75 and a precision of 0.62. While SMOTE, the model accuracy became 88%, with a precision value of 0.86, recall 0.87, and f1-score 0.87, which showed more balanced results across classes.