Implementation of the Double Exponential Smoothing Method for Pharmacy Empowerment through a Vitamin Sales Prediction System

Abstract

Vitamin sales at Pandu Jaya Pharmacy need to be accurately predicted to prevent overstocking or understocking, which can lead to losses. This study aims to design and develop a web-based application capable of predicting vitamin sales for the next four months as a basis for inventory management decisions. The data used are historical records of vitamin sales at Pandu Jaya Pharmacy over a certain time period, which are processed sequentially. The applied method is Double Exponential Smoothing (DES) with an optimization process of α (alpha) and β (beta) parameters to obtain the most accurate forecasting results. Model performance evaluation was carried out using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) metrics. The test results show that the DES model with optimal parameters produces a relatively small error value (RMSE = 61.96289), so that the prediction for the next four months shows a stable increasing sales trend. These findings indicate that the developed web-based prediction system can be an effective tool for pharmacy management in planning stock requirements, reducing the risk of losses, and supporting a more efficient procurement strategy.