• python 3.5
• CUDA 9.2 (optional, if set USE_GPU in config.py equal to True, you need a set up a complete cuda environment for tensorflow)

• tensorflow 1.10.1
• xgboost 0.80
• scikit-learn 0.19.1
• pandas 0.23.4
• numpy 1.15.1

The codes of data pre-process and feature enginnering are in preprocess floder. The result of feature engineering is also in Project5002.ipynb. Please move it to the floder containing original csv data files.

Before running main.py, you should run preprocess.py in preprocess firstly. Please copy original csv data files to the current floder (same as where the main.py is).

1. Edit the DATA_DIR and GEN_DATA_DIR in config.py.

2. (optional) Edit the other parameters in config.py

3. python main.py

Detailed code explanation please refer to the code comments.

https://github.com/fan31415/Attention-LSTM-on-Air-Quality-Prediction