<![CDATA[The advancement of information technology encourages businesses to utilize historical data for decision-making. CV Surya Perkasa, a paving block manufacturer, faces challenges in analyzing sales fluctuations due to the limited utilization of historical transaction data. This study aims to implement K-Nearest Neighbor (KNN) Regression as an initial simulation for estimating paving block sales volumes. The research employed a quantitative data mining approach using Python on Google Colab. The dataset comprised 500 sales transactions from 2022 to 2025, aggregated into monthly sales data by product type. Data preprocessing included cleaning, temporal aggregation, feature transformation, One-Hot Encoding, normalization, and historical feature construction using LAG_1 and LAG_12. Model performance was evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and R². The optimized model achieved an MAE of 331.36, MSE of 160,281.61, RMSE of 400.35, MAPE of 36.33%, and R² of 0.56. These results indicate that KNN Regression partially captures historical sales patterns but remains limited by demand fluctuations and historical features. Therefore, further improvements are required before operational forecasting implementation.]]>
Copyrights © 2026
