Analisis Safety Stock Produk Berbasis Pada Model Prediksi Penjualan Produk Coldmix Di PT XYZ

Abstract

PT XYZ merupakan perusahaan yang bergerak di bidang pemeliharaan infrastruktur jalan tol melalui unit Asphalt Mixing Plant (AMP). Salah satu permasalahan yang dihadapi adalah belum adanya sistem peramalan kuantitatif untuk memprediksi kebutuhan produk Coldmix, sehingga perencanaan persediaan kurang optimal di tengah permintaan yang fluktuatif. Penelitian ini bertujuan menganalisis pola historis permintaan dan penjualan Coldmix, mengembangkan model peramalan menggunakan Adaptive Neuro-Fuzzy Inference System (ANFIS), serta menentukan kebijakan persediaan optimal berupa safety stock dan reorder point. Hasil penelitian menunjukkan bahwa model ANFIS mampu memberikan hasil prediksi yang baik dengan nilai RMSE 21,0019, MAPE 14,6013%, R² 0,8997, dan korelasi 0,9485. Berdasarkan hasil prediksi tersebut diperoleh safety stock sebesar 2.032,12 zak dan reorder point sebesar 3.042,64 zak, sehingga integrasi ANFIS dengan kebijakan persediaan probabilistik mampu menghasilkan sistem perencanaan persediaan yang lebih adaptif dan efisien.

PT XYZ is a company engaged in toll road infrastructure maintenance through its Asphalt Mixing Plant (AMP) unit. One of the main challenges faced by the AMP unit is the absence of a quantitative forecasting system to predict Coldmix product demand, resulting in less optimal inventory planning amid fluctuating demand patterns. This study aims to analyze the historical demand and sales patterns of Coldmix, develop a forecasting model using the Adaptive Neuro-Fuzzy Inference System (ANFIS), and determine optimal inventory policies in the form of safety stock and reorder point. The results indicate that the ANFIS model provides good forecasting performance, with an RMSE of 21.0019, a MAPE of 14.6013%, an R² of 0.8997, and a correlation coefficient of 0.9485. Based on the forecasting results, the calculated safety stock and reorder point were 2,032.12 sacks and 3,042.64 sacks, respectively. Therefore, the integration of ANFIS with a probabilistic inventory policy is capable of producing a more adaptive and efficient inventory planning system.