<![CDATA[Fluctuations in pet shop product demand often cause inventory imbalances due to procurement planning based on subjective manager estimates. This study aims to evaluate the performance of the Multiple Linear Regression (MLR) algorithm in predicting daily sales by integrating time trends and temporal factors (weekly and monthly cycles). The methodology follows the CRISP-DM framework using transaction data from Cipanas Pets Shop. Partial hypothesis testing (t-test) revealed that the weekend variable () and sales trend () significantly increased daily sales. Conversely, the pay cycle was statistically insignificant (), confirming that pet shop products are classified as basic needs. Evaluation on 35% testing data yielded an MAE of 4.6146 units, RMSE of 5.4679 units, MAPE of 30.18% (feasible category), and of 0.1399. These findings were converted into a Continuous Review System formulation (Safety Stock and Reorder Point) and deployed into a web-based decision support system. This research provides a practical contribution to pet shop managers transitioning toward data-driven inventory management to mitigate overstocking and stockouts.]]>
Copyrights © 2026
