This project aims to predict medical insurance costs using regression techniques from supervised learning. All code and explanations are provided in the medical-insurance-cost-predict.ipynb notebook.
After thorough data exploration and preprocessing, various regression algorithms were tested and compared, including:
- Linear Regression (Benchmark Model)
- Decision Trees
- Random Forest
- Support Vector Machines
- K-Nearest Neighbors
Each model was evaluated based on key metrics like mean squared error (MSE) and R² score.
Among the algorithms, the Random Forest Regressor achieved the best performance, with the following metrics:
- MSE: 15.78%
- R² Score: 79.7%
This was a significant improvement over the benchmark model, Linear Regression, which had:
- MSE: 17.67%
- R² Score: 72.02%
Based on these results, the Random Forest model was selected, and further efforts were focused on optimizing this model.
-
Randomized Search: A random search was conducted to identify better hyperparameters, improving the model performance to:
- MSE: 13.74%
- R² Score: 81.41%
-
Grid Search with Cross-Validation: Further tuning with grid search provided a slight improvement in MSE but no significant changes overall.
The final optimized Random Forest model achieved:
- Mean Squared Error: 0.1373
- R² Score: 0.8142
This result demonstrates the effectiveness of the Random Forest approach for predicting medical insurance costs.