Implementasi Algoritma Machine Learning untuk Forecasting Demand Pada Usaha Kerupuk Sehat Krusawi

Abstract

The rapid development of information technology has encouraged business actors to utilize data analysis to improve efficiency and competitiveness, one of which is through demand forecasting. This study aims to implement machine learning algorithms to forecast product demand in the Krusawi Healthy Crackers business. The method employed is Prophet, which was selected due to its capability to handle time series data with nonlinear trends and seasonal patterns. The data used consist of historical daily sales data from April to July 2024, which were subsequently aggregated into weekly data. The research stages include data collection, data preprocessing (data aggregation, handling missing values, and Box-Cox transformation), Prophet model design with logistic growth and custom bi-monthly seasonality, model training, and performance evaluation. The results indicate that the Prophet model provides excellent forecasting performance, achieving a Mean Absolute Percentage Error (MAPE) of 6.57% or an accuracy level of 93.43%. The model successfully captures trend and seasonal patterns in Krusawi product sales. Therefore, the implementation of machine learning algorithms using the Prophet method proves to be a reliable solution for supporting production planning and inventory management in the Krusawi healthy crackers business, and has the potential to improve operational efficiency and business decision-making.