Estimation of Air Quality Index in Delhi with Neural Networks and PCA

Authors: Shriniketan Kulkarni, Harneet Singh Bali, and Rajashree Krishna

Source: https://ieeexplore.ieee.org/document/10133846

Overview:

This paper proposes a novel method for estimating the Air Quality Index (AQI) in Delhi, India, using a combination of principal component analysis (PCA), multiple regression, and artificial neural networks (ANNs). It aims to improve the accuracy of AQI prediction compared to traditional methods.

Key Points:

Data: Historical air quality and meteorological data from Delhi, including ozone, particulate matter (PM2.5 and PM10), sulfur dioxide, nitrogen dioxide, carbon monoxide, temperature, humidity, wind speed, and wind direction. Methodology: PCA: Identifies the most significant factors influencing AQI and reduces data dimensionality. Multiple Regression: Analyzes the identified factors and their impact on AQI. ANN: Trained on the processed data to learn complex relationships between input variables and AQI. Results: The proposed method outperforms traditional methods in AQI prediction accuracy. ANN shows superior performance compared to multiple regression alone. Potential Applications: Improved air quality forecasting in Delhi. Planning and implementing effective air pollution control measures. Providing real-time air quality information to the public. Getting Started:

Download the full paper from the provided link.

Install necessary libraries for data analysis and modeling (e.g., pandas, scikit-learn, tensorflow). Follow the steps outlined in the paper to reproduce the analysis and build your own AQI prediction model. Further Exploration:

Investigate the influence of different ANN architectures on prediction accuracy. Apply the model to other cities facing air quality challenges. Integrate the model with air quality monitoring systems for real-time predictions. Disclaimer:

This readme file is based on the available information about the paper. Please refer to the full paper for detailed methodology, results, and discussion.

We hope thisreadme.md file provides a helpful introduction to the paper and its contributions. Feel free to contact the authors or raise any questions you might have.

Note:

This research paper is publicly available on IEEE Xplore. For more information about the data or code used in this research, please contact one of the authors.

Code and Data Access: The code and other related materials are stored in a private repository. If you wish to access the codebase or have inquiries about the data, please reach out to one of the authors for further assistance.

We appreciate your interest in our work!