<![CDATA[Effective drug stock management is essential in healthcare services, particularly for infectious diseases such as pulmonary tuberculosis. This study aims to forecast TB drug stock using the ARIMA model and analyze residual volatility based on data from Banda Sakti Public Health Center, Lhokseumawe City. It focuses on applying predictive modeling at the primary healthcare level, which has rarely been addressed in previous studies. The dataset covers six drug types from January 2021 to December 2024. ARIMA models were selected automatically using Python and evaluated using sMAPE, MAE, and RMSE. Results show that ARIMA was successfully applied to four drug types, with sMAPE ranging from 31% to 41%, which is acceptable for short-term planning. The ARCH test produced p-values > 0.05, indicating that GARCH was not necessary. Two drug types could not be modeled due to zero-constant and sporadic data patterns. The proposed system can assist pharmacy staff in planning procurement and safety stock at the primary care level.]]>
Copyrights © 2025
