Pengembangan Model Prediksi Penjualan Mingguan pada Perusahaan Ritel Walmart Berbasis Artificial Neural Network

Development of a Weekly Sales Prediction Model at Walmart Retail Company Based on Artificial Neural Network

Abstract

Sales prediction is a crucial element in the modern retail industry, as it directly influences decision-making in inventory management, marketing strategy planning, and product distribution. This study aims to develop a weekly sales prediction model for Walmart stores using the Artificial Neural Network approach. The dataset comprises 5,508 historical records that have been cleaned from outliers and includes several variables such as temperature, fuel price, Consumer Price Index, unemployment rate, and holiday indicators. The model is designed using a Sequential architecture with hyperparameter optimization involving the number of neurons and learning rate. Model performance is evaluated using the Mean Squared Error metric.