Health Insurance Cost Prediction: Linear Regression Analysis

Project Overview

This project uses Linear Regression to predict individual medical insurance charges. By analyzing factors such as age, BMI, and smoking status, I developed a model that explains 99% of the variance in insurance costs, providing a clear mathematical framework for risk-based pricing.

Data Engineering & Modeling

• One-Hot Encoding: Transformed categorical variables (Smoker status, Region) into numerical values using pd.get_dummies while avoiding the Dummy Variable Trap.
• Algorithm: Implemented Scikit-Learn's LinearRegression to identify the specific "weights" of health risk factors.

Business Insights (The Coefficients)

My model identified the exact cost drivers for insurance premiums:

1. The Smoker Tax: Being a smoker increases charges by approximately $19,799.
2. The Aging Factor: For every single year of age, the premium increases by $248.64.
3. The BMI Impact: Each point of BMI adds roughly $95.21 to the annual cost.

Model Performance

• R-Squared Score: 0.99 (The model captures 99% of pricing logic).
• Mean Absolute Error (MAE): $806.51 (High precision in cost estimation).

Files in this Repo

• insurance_regression.ipynb: Complete code for encoding, training, and evaluation.
• actual_vs_predicted.png: Visualization confirming the model's accuracy.