Air Quality Index Prediction

ADNNet: Attention-based deep neural network for Air Quality Index prediction

Introduction

This repository provides the implementation of ADNNet for AQI prediction. The experiments have been performed datasets: AQI Online Testing and Analysis Platform.

If you use this repo, please cite our paper:

@article{wu2024adnnet,
  title={ADNNet: Attention-based deep neural network for Air Quality Index prediction},
  author={Wu, Xiankui and Gu, Xinyu and See, KW},
  journal={Expert Systems with Applications},
  pages={125128},
  year={2024},
  publisher={Elsevier}
}

Getting Started

(1) Install requirements

Python packages

Tested with TensorFlow2.8+Keras2.8.0.

pip install tensorflow
pip install pandas sklearn scipy
pip install plotly
pip install jupyter notebook ipykernel jupyterlab

(2) Download the datasets

Download the AQI Dataset and put its content in the directory AQI-Attention-based-DNN/data/

Test With Different Models

ADNNet.ipynb and ADNNet_multisteps.ipynb represent the implementation of one-step prediction and multi-step prediction respectively. --baseline: LSTM, N-BEATS, Informer, Autoformer

ADNNet.ipynb --feature=0 --model=0  

ADNNet_multisteps.ipynb --feature=0 --model=0