Demand Prediction Architecture for Distribution Business by Adopting Multiple Recurrent Neural Networks
-
- OKADOME Yuya
- HItachi, Ltd.
-
- WEI Wenpeng
- HItachi, Ltd.
-
- AIZONO Toshiko
- HItachi, Ltd.
Bibliographic Information
- Other Title
-
- 複数の再帰型ニューラルネットワークを用いた需要予測アーキテクチャの開発
Abstract
Due to change of market, an explosive variety of handled items brings considerable costs of both over-stocking and under-stocking to distributors. In this research, we propose the demand prediction architecture by adopting multiple recurrent neural network (RNN). The proposed model can handle various types of information, e.g., weather, store sales, by placing the RNNs at the input layer. We applied preposed model to a demand prediction problem using the open data which has the daily demands of 4,000 items. Results shows our proposed model achieves the accurate short- and long-term demand prediction
Journal
-
- 電子情報通信学会論文誌D 情報・システム
-
電子情報通信学会論文誌D 情報・システム J103-D (1), 24-33, 2020-01-01
The Institute of Electronics, Information and Communication Engineers
- Tweet
Details 詳細情報について
-
- CRID
- 1390846609785633152
-
- ISSN
- 18810225
- 18804535
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed