Analisis dan Implementasi Sistem Rekomendasi Tanaman Menggunakan Algoritma XGBoost dan CatBoost dengan Explainable AI (SHAP)

Abstrak

The agricultural sector faces serious challenges due to climate change and land degradation, necessitating a data-driven approach through precision agriculture. This study aims to develop an accurate and transparent crop recommendation system by comparing the XGBoost and CatBoost algorithms. This research is a quantitative study with a computational experimental design focusing on model development and evaluation. The dataset used is the Crop Recommendation Dataset consisting of 2,200 samples, which were processed using label encoding and evaluated via 5-Fold Cross Validation. To enhance interpretability, the SHAP method was applied to explain the contribution of each feature to the model's predictions. The results show that both models achieve exceptionally high performance, with accuracies exceeding 99%. CatBoost recorded the highest accuracy in a single test at 99.32%, but XGBoost demonstrated superiority in stability and generalization capability with an average accuracy of 99.45% and the lowest standard deviation ($\pm$0.0023). SHAP analysis revealed that rainfall and nutrient factors were the most influential variables, aligning with agronomic principles. The XGBoost model was then implemented into a web-based system, producing outputs consistent with manual testing. This study concludes that the integration of machine learning and Explainable AI (XAI) can deliver an accurate, stable, and trustworthy recommendation system to support decision-making in agriculture.

Identifier

• OpenAlex: https://openalex.org/W7164684640
• DOI: https://doi.org/10.54373/hijm.v4i3.5546