🌍 India AQI Analysis & Prediction (Machine Learning)

📌 Overview

This project focuses on analyzing air quality data across multiple cities in India and predicting PM2.5 levels using Machine Learning techniques.

Air pollution is a major environmental issue affecting human health. This project uses data-driven approaches to understand pollution patterns and build predictive models for air quality monitoring.

---

🎯 Objectives

• Analyze air quality data across Indian cities
• Perform data preprocessing and transformation
• Identify pollution trends using visualization
• Build machine learning models to predict PM2.5
• Compare model performance and select the best model

---

📊 Dataset

The dataset contains air pollution data collected from various cities and monitoring stations.

Features:

• PM2.5 (Target Variable)
• PM10
• CO
• NO2
• SO2
• OZONE

Note: NH3 was excluded from the dataset due to a high number of missing values to ensure better model performance and data reliability.

---

⚙️ Workflow

1. Data Collection

2. Data Preprocessing

   • Removed irrelevant columns
   • Handled missing values
   • Applied pivot table transformation

3. Exploratory Data Analysis (EDA)

4. Train-Test Split (80:20)

5. Model Training

6. Evaluation and Comparison

---

🤖 Machine Learning Models Used

1. Linear Regression

• Assumes linear relationship between pollutants and PM2.5
• Provided best performance

2. Decision Tree

• Captures non-linear relationships
• Prone to overfitting

3. Random Forest

• Ensemble method (multiple decision trees)
• Better than Decision Tree but not best

---

📈 Evaluation Metrics

• MAE (Mean Absolute Error)
• RMSE (Root Mean Squared Error)

---

📊 Results

Model	MAE	RMSE	Performance
Linear Regression	15.66	20.49	⭐ Best
Decision Tree	17.71	25.78	Overfitting
Random Forest	16.66	22.58	Moderate

🔍 Key Insight:

Linear Regression performed best because the dataset shows a linear relationship between pollutants and PM2.5.

---

📊 Key Observations

• PM2.5 is strongly influenced by PM10, NO2, and CO
• Some cities show significantly higher pollution levels
• Pollution distribution varies across regions
• Most values fall in moderate pollution range with some extreme cases

---

🛠 Tech Stack

• Python
• Pandas
• NumPy
• Matplotlib
• Seaborn
• Scikit-learn

---

📁 Project Structure

AQI-ML-Prediction/ │ ├── AQI_ML_Prediction.ipynb ├── aqi2.csv ├── README.md

---

🚀 Conclusion

This project demonstrates how Machine Learning can be used to analyze environmental data and predict air pollution levels effectively.

The results highlight that simpler models like Linear Regression can outperform complex models depending on the dataset characteristics.

---

🔗 Related Project

📊 Power BI Dashboard: https://github.com/the-om-shinde/AQI-Analysis-Dashboard

---

📚 References

• Scikit-learn Documentation
• Pandas Documentation
• Kaggle Air Quality Dataset
• Machine Learning Books