Abstract
It is essential that managers have information on how to design bundling incentives to increase sales for the worst selling products. Therefore, this study attempts to identify which items are suitable for promoting the worst selling products by different measures and further predict the sales of products by using deep learning approaches, recurrent neural network and long short-term memory (RNN and LSTM), as well as two conventional methods, linear regression and XGBRegressor. Experimental results from a real-life dataset show that the proposed measure, PromoteRate, outperforms the other traditional measures, lift, cosine, and allconf. That is, PromoteRate can discover better bundling promotion items for the worst selling products. LSTM shows superior predictive performance compared to other approaches. Finally, the experiment also includes results for products with lower sales, not solely focused on the worst performers.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call