Enhancing Sales Forecasting: Leveraging Retail Sales Data for Advanced AI Predictive Models

Summary

This thesis aims to improve sales forecasting in the retail sector through the application of advanced Artificial Intelligence (AI) techniques. It addresses the challenge of fluctuations in retail sales, which are influenced by various external factors, including economic changes and shifts in consumer behavior. The study develops and evaluates multiple AI forecasting models, such as FBProphet, NeuralProphet, XGBoost, LSTM, TFT, and TimeGPT, to enhance the accuracy and flexibility of predictions. Furthermore, it provides an in-depth comparison between conventional time series forecasting methods, such as ARIMA, and the aforementioned machine learning and deep learning approaches. The findings highlight the superior performance of state-of-the-art AI-based models in handling complex patterns and adapting to new data, thereby offering more accurate sales forecasts. Additionally, the thesis underscores the importance and impact of thorough data preprocessing.