Penerapan Metode Algoritma Extreme Gradient Boosting Dalam Memprediksi Penjualan Thrifting Pada Toko Cchase

Abstrak

Thrifting nowadays is very easy to access since thrift stores have been rapidly emerging in urban areas. However, one of the common problems faced by thrift shops is related to stock management and sales prediction. For example, at Cchase store, stock estimation is often inaccurate due to the wide variety of thrift items and the uncertain supply, which leads to the risk of unsold inventory. To overcome this issue, this study applies machine learning technology using the Extreme Gradient Boosting (XGBoost) algorithm. This method was chosen because it has been proven to provide accurate prediction results on data with complex patterns. The data used consist of one year of sales records, which were processed, split into training and testing sets, and then evaluated using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) metrics. The results show that the model achieved an RMSE of 0.8105 and an MAE of 0.6643, indicating that the model performs well in predicting sales. Furthermore, the prediction results for the upcoming month reveal the top three product categories with the highest sales, namely crewnecks, hoodies, and t-shirts. These findings are expected to help thrift business owners manage stock more efficiently and develop more effective sales strategies.