Air Quality Regression: Linear Models vs Decision Trees
CO prediction on the UCI Air Quality dataset: linear regression against decision trees tuned with randomised search.
What was built
Exploratory analysis finds missing values coded as -200 and compares possible targets. I chose CO(GT). Simple and multiple linear regression are compared with decision-tree regressors. A tree tuned with RandomizedSearchCV (400 candidates, 5-fold) reaches a test R-squared of 0.843.
Facts
| Item | Value |
|---|---|
| Tuned tree, test R-squared | 0.843 |
| Dataset rows | 9,471 |
Links
No public repository for this project.