Evaluating the efficacy of gradient boosting algorithms in retail demand forecasting: a case study of triumph international

Abstract

This thesis is the result of my internship at The Data Cooks, an Amsterdam-based data agency, and aims to enhance demand forecasting for Triumph International, a leading lingerie manufacturer. The research is focused on addressing the complex challenge of disaggregated forecasting in the fashion retail industry, often characterized by a large number of stores and products. Central to this research is the application of gradient boosting algorithms, a cutting edge approach in machine learning. By focusing on methods such as LightGBM and XGBoost, this study delves into the effectiveness of ensemble learning in handling complex, time sensitive data.