Introducing
Your new presentation assistant.
Refine, enhance, and tailor your content, source relevant images, and edit visuals quicker than ever before.
Trending searches
Air Pollution Prediction using Machine Learning
13
Air Pollution Prediction using Machine Learning
Presented by:
Melanie Chloe
Jeetika Ramgoolam
Kaviswaree Nirsimloo-Ramasawmy
Our presentation is based on a research done by the International Research Journal of Engineering and Technology in 2019 where air pollution is predicted using machine learning.
WHO has measured quality of air in approximately 1500 cities and Indian capital city was the one of the most polluted cities around the world. Pune is having highest concentration of particulate matter which is smaller than 2.5 micrometre.
Introduction
Different sources of air pollution
Construction and Demolition
Pavement, road dust
and construction activities in Pune are considered to be a major contributor of increasing particulate pollution.
Different sources of air pollution
Vehicular Emission
Due to increase in vehicular traffic that includes commercial vehicles, vehicles with gears and without gears and heavy load vehicles in Pune, leads to increase the pollution in the air and vehicles are considered
to be the major cause of air pollution.
Industrial Emission
As the new technologies are evolving day by day so new factories are being installed. This has caused air pollution by emitting harmful smoke, gases etc.
The study area
The study area
Pune is one of the most polluted cities in India. Peak levels of fine particulate matter (PM) in Pune increased by about 70%, basically due to high industrial and vehicular emissions, construction work and crop burning. The level of the airborne particulate matter- PM2.5 is very high in Pune. It is considered to be the most harmful pollutants to health.
A lot of work is done in the study and analysis of
air pollution as well as predicting the future trends. Three different methods of predictions have been done
Study and analysis of
air pollution
Linear regression
Linear regression-based air pollution
Research 1
Nitin Sadashiv Desai, IoT based air pollution monitoring
and predictor system on Beagle Bone Black, IEEE-2017,
978-1-5090-5913-3/17/531.00, March 2017.
In resaerch 1, Linear regression-based air pollution prediction was done. It suggests cloud data for data analytics which can be used for taking the decision to minimize pollution. But they have used BI service and Microsoft Azure for analysis which is very expensive services. The model is not very accurate because of linear regression-based model.
Machine learning
Machine learning
In research 2, machine learning based air pollution prediction is done. It suggests multilayer perceptron which results in very accurate result. But it takes large datasets and long duration
for training.
Research 2
Kettun Oberoi, Predicting Trends in Air Pollution in
Delhi using Data Mining, IEEE-2016, 978-1-4673-6984-
8/16/531.00, January 2016.M. Young, The Technical
Writer’s Handbook. Mill Valley, CA: University Science,
1989.
Recurrent Neural Network
Recurrent Neural Network based model
Research 3
Yue Shan Chang, Big data platform for air quality
analysis and prediction, IEEE-2018, 978-5636-4959-
6/18/531.00, February 2018.
In research 3 Recurrent Neural Network based model for air
pollution prediction is done. It suggests using machine
learning algorithm and recurrent neural network for
prediction which generates most accurate result but, its very
expensive to implement.
Proposed approach
Proposed approach
Data set used
Data set
Data from Central Pollution Control Board (CPCB) has been taken. This data set consists of attributes that are time (in months), air pollutants like SO2, NO2, CO, PM10 and Ozone (O3). Data is collected from 2000 to 2018 to predict the trends of the above pollutants in upcoming years.
One example of the data set
Flow chart of
proposed work
After collecting the data, it is preprocessed. In preprocessing data are cleaned by removing noise and filling up the missing values. Multilayer perception is used for prediction and presented to the user on the app.
Flow chart of proposed work
Flowchart of proposed work
Multilayer perceptron
Multilayer perceptron
It is a class of feed forward artificial neural network. It
consists of at least, three layers of nodes: an input layer, a hidden layer and an output layer. Except for the input nodes, each node is a neuron that uses a nonlinear activation function. MLP utilizes a supervised learning technique called back- propagation for training. Its multiple layers and nonlinear activation distinguish MLP from a linear perceptron.