Air Quality Index (AQI) Multiclass Classification :

This repository contains the implementation of the Machine learning project which consists of 3 major parts :

1. Performing exploratory data analysis on a chosen dataset, including data wrangling, cleaning etc.
2. Implementing different classification algortihms and evaluating their perofrmance.
3. Benchmarking the classification results and interpreting them.

Dataset :

I chose the World Air Quality Index by City and Coordinates on Kaggle

(click here to obtain it : https://www.kaggle.com/datasets/adityaramachandran27/world-air-quality-index-by-city-and-coordinates )

Exploratory Data Analysis :

• Understadning the structure of the data
• Handling missing and duplicated values
• Understanding the data distribution
• Exploring categorical values and encoding them
• Data visualization
• Feature Engineering
• Pollars alternative

Classification Algorithms :

I chose 11 different classification algortihms. In each one, I experimented with different techniques to enhance the results of the classification, like normalizing and scaling the data, performing hyperparamter tuning, using resampling techhniques to mitigate class imbalance etc.

1. K-Nearest Neighbors

2. Logistic regression

3. Decision Trees

4. Random Forests

5. Support Vector Machine

6. Naive Bayes

7. Neural Networks

8. Gradient Boosting Machine

9. Linear Descrimant Analysis

10. XGBoost

Benchmarking results

Algorithms	Accuracy	Recall	Precision	F1-score
K-Nearest-Neighbour	0.9902	0.9902	0.9908	0.9891
Decision Trees	0.9979	0.9979	0.9983	0.9978
Random Forests	0.9972	0.9961	0.9971	0.9961
Naive Bayes	0.4570	0.4570	0.6317	0.4009
Support Vector Machine	0.9926	0.9926	0.9928	0.9968
Logistic Regression	0.9740	0.9740	0.9753	0.9703
Neural Networks	0.8507	0.8507	0.8726	0.8442
Gradient Boosting Machine	0.9968	0.9968	0.9973	0.9967
Linear Descriminant Analysis	0.9796	0.9796	0.9805	0.9793
XGBoost	0.9982	0.9982	0.9985	0.9980
AdaBoost	0.9965	0.9965	0.9973	0.9964