Goal : Developed a machine learning model to accurately predict air pollution levels based on environmental and meteorological features.
What Was Done
Performed data preprocessing, including missing value imputation, duplicate removal, outlier handling, and label encoding.
Conducted exploratory data analysis (EDA) using statistical visualizations and correlation analysis.
Trained and compared multiple classification models: Logistic Regression, Decision Tree, Gaussian Naive Bayes, and Random Forest.
Evaluated model performance using Accuracy, Confusion Matrix, and Classification Report.
Built a Scikit-learn Pipeline and deployed the best-performing Random Forest model using Pickle.
Results
Random Forest (Best Model): 94.86% Accuracy
Gaussian Naive Bayes: 92.58%
Decision Tree: 88.02%
Logistic Regression: 68.90%
Tech Stack
Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Pickle
