Pengembangan Sistem Rekomendasi Order Barang Menggunakan Algoritma Random Forest dengan Metode CRISP-DM

Abstract

PT XYZ mengalami permasalahan pengelolaan stok karena sistem rekomendasi order masih menggunakan perhitungan statis sehingga belum mampu menangkap pola permintaan secara dinamis. Penelitian ini bertujuan mengembangkan sistem rekomendasi order barang menggunakan algoritma Random Forest dengan metode CRISP-DM. Pada tahapan pembuatan model dan evaluasi dilakukan secara iteratif melalui tiga skenario model, yaitu baseline, clipping, dan log transformation. Evaluasi model dilakukan menggunakan metrik Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Coefficient of Determination (R²), dan Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan bahwa model baseline menghasilkan performa terbaik dengan nilai MAE sebesar 1,36, RMSE sebesar 68,88, R² sebesar 0,84, dan MAPE sebesar 35,67%. Hasil prediksi permintaan kemudian diimplementasikan pada sistem berbasis Python yang terintegrasi dengan website order untuk mendukung proses rekomendasi pemesanan barang secara lebih adaptif dan optimal.

PT XYZ experienced inventory management problems because the order recommendation system still relied on static calculations and was unable to capture demand patterns dynamically. This study aimed to develop an item order recommendation system using the Random Forest algorithm with the CRISP-DM method. The modeling and evaluation stages were carried out iteratively through three model scenarios: baseline, clipping, and log transformation. Model evaluation was conducted using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Coefficient of Determination (R²), and Mean Absolute Percentage Error (MAPE). The results showed that the baseline model achieved the best performance with an MAE of 1,36, RMSE of 68,88, R² of 0,84, and MAPE of 35,67%. The demand prediction results were then implemented into a Python-based system integrated with the order website to support a more adaptive and optimal item ordering recommendation process.