An artificial neural network for mixed frequency data

作者：

Highlights：

• The ANN-(U-)MIDAS model is developed.

• It models artificial neural network on mixed frequency data via (U-)MIDAS.

• It can be estimated by the standard gradient based optimization algorithm.

• It is flexible to explore potential nonlinear patterns among variables.

• It has been successfully applied to better Chinese inflation forecasts.

摘要

•The ANN-(U-)MIDAS model is developed.•It models artificial neural network on mixed frequency data via (U-)MIDAS.•It can be estimated by the standard gradient based optimization algorithm.•It is flexible to explore potential nonlinear patterns among variables.•It has been successfully applied to better Chinese inflation forecasts.

论文关键词：Artificial neural network,Mixed frequency data,Mixed data sampling (MIDAS),Nonlinear pattern,Inflation forecasting

论文评审过程：Received 11 July 2018, Revised 3 October 2018, Accepted 6 October 2018, Available online 9 October 2018, Version of Record 11 October 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.10.013