A Hybrid Machine Learning Approach for Predicting Product Life Span and Total Lifetime Sales

Predicting and managing a product's life cycle is essential for enhancing a company's competitiveness. Accurate prediction of a product's total sales volume over its lifespan enables companies to make informed decisions about product planning and sales strategies, leading to improved effectiveness. The objective of this study is to develop a precise hybrid machine-learning model for predicting the total sales volume of a product. To achieve our goal, we propose a hybrid model that combines LGBM (Light Gradient-Boosting Machine) and QBoost (Quantile Boost). LGBM is employed to predict the lifespan of the product, while QBoost utilizes a dataset that includes product lifespan information to predict the total demand over the product's entire lifetime. We assess the performance of our proposed model using four evaluation metrics: RMSE, NRMSE, MAE, and mMAPE. We conduct a comparative analysis of our hybrid model with single models, including Random Forest, LGBM, and QBoost. Additionally, we evaluate the performance of another hybrid model that combines these individual models. The analysis unequivocally demonstrates that the proposed model surpasses benchmark models across all evaluation metrics. This study has significant implications for improving decision-making in product lifecycle management through accurate prediction of product life expectancy and total lifetime sales volume.