Model Prediksi Data Besar Distribusi Produk Farmasi: Analisis Kinerja Model Deep

Abstract

As the company's business goes on, the problems in storing and processing big data will become more complex. Unorganized data can cause companies to fail in maximizing sales strategies. One approach to maximize the sales strategy is forecasting. This study aims to reduce short-term customer inventory levels and assist in determining realistic sales targets in the future by proposing a deep learning method based on customer segmentation. The analytical framework is proposed using the Robust Principal Component Analysis (RPCA) technique to reduce the dimensions of the dataset, then the K-Means Clustering algorithm is applied to identify population groups in order to see several clusters that can best represent the characteristics of the company's existing customer base. Finally, the CNN and LSTM layers are combined to estimate future sales. Forecasting results were evaluated using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The proposed approach is to fill the gaps in problems that occur due to lack of information regarding the lack of information about business performance in terms of product categorization