Approaching sales forecasting using recurrent neural networks and transformers

作者：

Highlights：

• Deep learning algorithms achieve competitive results in sales forecast.

• A single model is needed to generalize over all products, stores and time.

• Random max time step trick can be used to avoid overfitting over specific timesteps.

摘要

•Deep learning algorithms achieve competitive results in sales forecast.•A single model is needed to generalize over all products, stores and time.•Random max time step trick can be used to avoid overfitting over specific timesteps.

论文关键词：Sales forecast,Supply chain,Deep learning,Transformer,Sequence to sequence

论文评审过程：Received 6 May 2021, Revised 30 December 2021, Accepted 25 March 2022, Available online 6 April 2022, Version of Record 25 April 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.116993