Research Article
Sales Prediction Based on State-of-art Machine Learning Scenarios
@INPROCEEDINGS{10.4108/eai.28-10-2022.2328458, author={Yuanpu Hu}, title={Sales Prediction Based on State-of-art Machine Learning Scenarios}, proceedings={Proceedings of the International Conference on Financial Innovation, FinTech and Information Technology, FFIT 2022, October 28-30, 2022, Shenzhen, China}, publisher={EAI}, proceedings_a={FFIT}, year={2023}, month={4}, keywords={sales forecast; machine learning; combination model; feature construction; sample weighting}, doi={10.4108/eai.28-10-2022.2328458} }- Yuanpu Hu
Year: 2023
Sales Prediction Based on State-of-art Machine Learning Scenarios
FFIT
EAI
DOI: 10.4108/eai.28-10-2022.2328458
Abstract
The way to predict sales volume accurately and efficiently is an important issue that enterprises have always paid attention to. Although the traditional time series prediction method plays a leading role in research and practice, it has some limitations. With the development of big data, e-commerce enterprises can obtain unprecedented data volume and data characteristics. It is difficult to accurately predict sales volume only based on past behaviors and trends. In this paper, we propose a combined forecasting model of cost aversion bias based on random forest, GBDT and XGboost algorithms, and uses the cost data of each commodity to realize the fine weighting of samples, so as to output the forecasting results. These results shed light on pointing that combined forecasting model can predict sales more accurately, which is of great significance to e-commerce enterprises to reduce commodity management costs.
