Forecasting Pharmaceutical Sales Using the ARIMA Method

Abstract

The increasing complexity of the pharmaceutical industry necessitates adaptive and predictive inventory management systems to mitigate the risks of stockout and cost inefficiency. This study aims to enhance the accuracy of drug demand forecasting at PT. XYZ, a vital pharmaceutical distributor, through the implementation of the Time Series Model ARIMA (Autoregressive Integrated Moving Average). A historical dataset of 17,521 drug sales transactions was processed and modeled to project future demand.Empirical results demonstrate that the ARIMA (1,1,1) model proves to be a valid and robust pharmacoeconomic instrument, achieving a superior level of accuracy (MSE Accuracy, 91.20% and RMSE Accuracy. 92.50%). These findings underscore the model's capability in predicting the time series pattern of drug sales with minimal error.The primary implication of this forecasting lies in pharmaceutical supply chain management. The analysis identifies distinct strategic needs Essential Fast-Moving products (e.g., Neozep Forte), which exhibit an upward growth trend ($GF=1.13$) and high Risk Score (1.49), require meticulous Safety Stock determination to ensure therapeutic availability (drug availability) and maintain the continuity of primary healthcare services. Conversely, highly volatile Slow-Moving products (e.g., Biogesic Anak, Risk Score 1.76) demand a minimal procurement strategy to reduce losses from overstocking and the potential for expired stock. Overall, the implementation of ARIMA supports data-driven logistical decision-making, enabling PT. XYZ to balance operational cost optimization with its professional pharmaceutical responsibility in guaranteeing a safe, precisely quantified, and efficient drug supply.