Machine Learning Techniques in Predicting Sales a Case Study of Jumia

The retail industry has experienced significant growth with the advent of e-commerce platforms like Jumia. Predicting sales accurately is critical for inventory management, marketing strategies, and overall operational efficiency. This paper explores the application of machine learning techniques to predict sales on Jumia, leveraging historical sales data and other relevant features to build and evaluate predictive models. Our results demonstrate that advanced machine learning models, particularly the gradient boosting machine, significantly outperform the baseline linear regression model. The gradient boosting machine achieved the lowest Mean Absolute Error (MAE) and Mean Squared Error (MSE), highlighting its superior prediction accuracy. Feature importance analysis revealed that pricing, promotional activities, and seasonal factors are key drivers of sales. These findings indicate that machine learning can effectively capture sales patterns, providing valuable insights for decision-makers.