💸 Health Insurance Cost Predictor

A machine learning project using Linear Regression to predict health insurance expenses based on personal and lifestyle data. Built using TensorFlow 2.x and trained on real-world data from insurance.csv.

---

📁 Dataset

The dataset contains the following features:

• age – Age of primary beneficiary
• sex – Gender (male, female)
• bmi – Body mass index
• children – Number of dependents
• smoker – Whether the person smokes (yes, no)
• region – Residential area in the US (northeast, northwest, etc.)
• expenses – Medical costs billed by health insurance

---

🔄 Preprocessing

• One-hot encoding applied to:
  • sex, smoker, and region (with drop_first=True to avoid dummy variable trap)
• expenses column popped as target variable
• Train-test split: 80% training / 20% testing
• StandardScaler used to normalize feature columns

---

🧠 Model Architecture

Built using TensorFlow Keras Sequential API:

• Dense(256) → ReLU
• Dropout(0.1)
• Dense(128) → ReLU
• Dropout(0.1)
• Dense(64) → ReLU
• Dense(1) → Output layer (regression)

Compiled with:

• Loss: Mean Squared Error (MSE)
• Optimizer: Adam
• Metrics: Mean Absolute Error (MAE)

EarlyStopping used to prevent overfitting.

---

📊 Results

• Evaluated on unseen test set
• Achieved MAE < 3500, passing the freeCodeCamp challenge ✅

Example output:

---

🚀 How to Run

1. Load the notebook in Google Colab
2. Run all cells (training will auto-start)
3. Final cell evaluates the model and displays predictions vs true values on a scatter plot

---

🧾 Challenge Objective

Train a regression model that can predict healthcare costs within a $3500 error margin on new, unseen data. Mission accomplished.

---