AQI Prediction.
A regression study that predicts the Air Quality Index, comparing Linear, Lasso and Ridge models.
On a Kaggle dataset of daily air-quality readings for Indian cities: gaps in lightly missing columns dropped and the rest mean-imputed, benzene, toluene and xylene turned into above-threshold flags, skewed pollutants Yeo-Johnson-transformed and the target log-scaled. After an 80/20 split, features are standard-scaled on the training set only, and Linear, Lasso and Ridge regression are compared on train and test MSE, RMSE and R², alongside exploratory histograms and a correlation heatmap.
- Role
- Solo project
- Time
- 2024
- Stack
- Python · scikit-learn · pandas · SciPy
- Access
- Source code
- Explore: A Kaggle dataset of daily air-quality readings for Indian cities, explored with histograms and, once transformed, a correlation heatmap.
- Prepare: Rows missing the AQI are dropped and other gaps dropped or mean-imputed; three solvents become threshold flags, skewed pollutants get a Yeo-Johnson transform, and the AQI target is log-scaled.
- Train: The rows are split 80/20. Features are standard-scaled on the training set only, then plain linear regression, Lasso and Ridge (which add regularisation) learn from the same training split.
- Compare: Each is scored on MSE, RMSE and R², on the training data and on the held-out 20%.
Explore. A Kaggle dataset of daily air-quality readings for Indian cities, explored with histograms and, once transformed, a correlation heatmap.
Stack
Python · scikit-learn · pandas · SciPy · matplotlib · seaborn
Want to talk about this build?
I'm open to remote full-time roles and freelance work, and happy to walk through the decisions behind it.