Description

There are my open-souce code by Keras according to the paper
XinLi,LidongBing,WaiLam and BeiShi. A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction

layer_definition

This part includes writing own layers.

dual_attention

This part includes data's preprocessing, model's building and model's training.

data

Downloaded from (https://cseweb.ucsd.edu/~yaq007/NASDAQ100_stock_data.html)