AirPol
Atharva Peshkar
Nagpur, Maharashtra
- 0 Collaborators
AirPol is a Machine Learning based project that aims to use various meteorological, geographical and temporal factors along with the current air pollutant concentration to accurately predict concentration of air pollutants in advance. ...learn more
Project status: Under Development
Intel Technologies
Intel Python
Overview / Usage
Currently there is no system to predict the Air Quality of a particular area well in advance so that precautions can be taken to avoid the entry of hazardous air pollutants in our biological system. My aim is to give the Air Quality Forecast in advance so that we can plan our day accordingly and avoid the harmful pollutants from entering our respiratory system.
Methodology / Approach
I am using the data provided by the government of India to train our Boosted Random Forest model that has been identified as the best performer among multiple estimator models. The data consists of about 300k data points. I have fine tuned the model for optimal performance.
This model will be deployed in the cloud. The required data will be collected by smartphones through the sensors in the smartphones and will be sent to the cloud where it will be processed for giving the predictions.
The current work has succesfully conducted a comparative study between various estimators to predict the air pollutant concentration. The algorithms considered during the comparative study were:
- Linear Regression
- Stochastic Gradient Descent
- Neural Network
- Decision Tree Regression
- Boosted Decision Tree Regression
- Random Forest
- Boosted Random Forest Regression
Other data analysis included splitting the dataset according to seasons, including and excluding meteorological factors etc.
Technologies Used
Technologies used include:
- Intel CPU
- Microsoft Windows 10
- Jupyter Notebook
- Intel Python
Libraries:
- Tensorflow = 2.x.x
- Scikit Learn
- Matplotlib
- Seaborn
- Numpy
- Pandas
- Scipy
Documents and Presentations
Repository
https://github.com/Atharva-Peshkar/AirPol-Air-pollutant-concentration-prediction-