AIR QUALITY

PREDICTION
Presented by:
P.Tejashwan (18E31A0538)
M.Chaitanya(18M31A0512)
FOR SMART T.Bhavana(18E31A0548)

CITIES
 CONTENTS
• Abstract
• Existing system
• Disadvantages
• Proposed system
• Advantages
• Dataset information
• Modules
• Hardware requirements
• Software requirements
• How to run the project?
• Conclusion
ABSTRACT:

• In recent years, people have been paying more and more attention to air quality. By using various
machine learning algorithms to predict air quality as this air quality put heavy effect on human
health.

• Aiming at this problem, we propose an improved air quality prediction method based on the
Light GBM model to predict the PM2.5 and PM10 concentration in the respective places.

• In this paper we are comparing RMSE (Root Mean Square Error) error rate with other
algorithms such as ADABOOST, GBDT (Gradient Boosting Decision Tree), XGBoost and DNN.

• The lower the RMSE error rate the better is the prediction model and in all algorithms Light
GBM is giving less error rate.
EXISTING SYSTEM:

• The Artificial Neural Network (ANN) modelling is presented by researchers for

• One such work is the steam gasification of palm kernel shell using CaO adsorbent
and coal bottom ash as a catalyst .

• The effect of the parameters such as temperature, CaO/biomass ratio and Coal
bottom ash weight percentage are modelled using ANN.
DISADVANTAGES:

• Predicting air quality is difficult

• Predicting temperature not possible

• Time taken process

• The proposed system aims at predicting the air quality using the Light GBM model.

• The model is trained with the statistical features of the historical air quality data and
meteorological data collected over the past three years.

• By providing the weather forecast data of any particular day and pollutant data of any nearby
day, we can predict the air quality of that day.

• An accuracy of 98.38% was obtained upon testing the model.

• Light GBM is found to be a suitable framework for prediction of air quality with high
accuracy, such that it is a better framework candidate for air quality prediction than the existing
models.
ADVANTAGES:

• Air quality prediction done easily by PM2.5 Quality

• Temperature predicted by PM10 Quality Prediction

 In above dataset screen we have city name, date and other air quality values and we will use
above dataset with above mention algorithms to predict air quality and compare RMSE error
rate.
MODULES:

1) Upload Air Quality Dataset: Using this module we will upload dataset to application and then
calculate features importance using correlation formula. The feature which is important for
prediction will have high value

2) Pre-process Dataset: using this module we will analyse entire dataset and then remove empty
and missing values

3) Run PM2.5 Quality Prediction: using this module we will predict PM2.5 air quality by using
Light GBM, GBDT, XGBOOST and DNN.

4) Run PM10 Quality Prediction: using this module we will predict PM10 air quality by using
Light GBM, GBDT, XGBOOST and DNN.

5) Comparison Graph: using this module we will compare RMSE error rate between all algorithms
for PM2.5 and PM10.
HARDWARE REQUIREMENTS:

• System : Pentium IV 2.4 GHz.

• Hard Disk : 40 GB.

• Floppy Drive : 1.44 Mb.

• Monitor : 15 VGA Colour.

• Mouse : Logitech.

• Ram : 512 Mb.

• Operating system : Windows 7 and above.

• Coding Language : python

• Frame work : tkinter

• In above screen click on ‘Upload Air Quality Dataset’ button to upload dataset and to get
below screen
• In above screen selecting and uploading ‘city_day.csv’ air quality dataset and then click
on ‘Open’ button to load dataset and to get below screen
• In above screen dataset loaded and we can see dataset values in text area and in graph we can see
column name from dataset and its correlation feature importance values in graph boxes. Now close
above graph and click on ‘Preprocess Dataset’ button to remove missing values and then calculate air
pollution rate date wise in dataset like below screen
• In above screen graph x-axis represents date and y-axis represents air pollution on that date. Now
close above graph to see pollution values
• In above screen for few days I am printing pollution rate and now all missing values are removed and
now click on ‘Run PM2.5 Prediction’ button to predict air quality and for testing purpose we are using
test values from dataset and then calculate RMSE error rate between original and prediction values
• In above black screen we can see predicted value and expected value as the original value
and we can see close difference between original (expected value) and predicted value
• In above screen we can see RMSE error rate for all 4 algorithms and in all algorithms Light GBM got
less RMSE error rate. Below is the predicted and original values from all 4 algorithms
• In above graph red line represents original air quality and green line represents predicted air quality and I
am displaying prediction graph for all 4 algorithms and from all 4 Light GBM is closed and now closed
above graph and then click on ‘ Run PM10 Quality Prediction’ button to predict air quality for PM10 using
all 4 algorithms
• In above screen all 4 algorithms run for air quality PM 10 and in all algorithms Light GBM
got less RMSE error and below is the prediction graph for all 4 algorithms
• Now click on ‘Comparison Graph’ button to get RMSE comparison graph for all 4
algorithms
• In above graph x-axis represents algorithm names and y-axis represents RMSE error rate and
for both PM10 and PM2.5 air quality Light GBM got less RMSE error
CONCLUSION:

By the test we performed it is clear that the algorithm Light BGM has the least root mean
square error value and it gives the accurate results than all the other algorithms and we
can also see the graph of the air quality and we can know what is the quality of air which
we are in taking daily and can take the precautions and remedies to avoid polluted air and
improve the quality of air.
ANY QUERIES?