Skip to content

DanielLin1986/DeepLearningApproachesForOzonePrediction

Repository files navigation

Predicting Ground-level Ozone Concentrations Using Deep Learning Networks

This repository contains Python code designed to evaluate six prominent deep learning (DL) architectures for predicting ground-level ozone concentrations. The architectures under examination include the Fully Connected Network (FCN), commonly referred to as the Multi-Layer Perceptron (MLP), and several variants Recurrent Neural Networks (RNNs), specifically the Long Short-Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM) models. Additionally, the Convolutional Neural Network (CNN) and a Transformer-based network are also involved. Moreover, to provide a comprehensive comparison, a conventional Machine Learning (ML) model—LightGBM is included. This allows for an evaluation of how DL approaches stack up against established ML methods, offering a robust framework for assessing the effectiveness of various architectures in predicting ozone concentrations. The implementation of DL models is based on Keras (version 2.6.0) with Tensorflow backend (version 2.6.0).

Run The Code

Please set up the environment based on the requirements.txt. Then run the .py file, for example:

Python OzonePrediction_2-Layer_Conv1d.py

For the complete dataset, please contact yfchi@fjsmu.edu.cn.

About

Applying different deep learning models for ground-level ozone concentration prediction.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages