A-Shahat/Egypt-Healthcare-Cost-Prediction

Predicting annual healthcare expenditures using demographic and clinical features — built with Python and Scikit-learn. # Egypt Healthcare Cost Prediction > Predicting annual healthcare expenditures using demographic and clinical features — built with Python and Scikit-learn. --- ## Overview Healthcare systems face increasing financial pressure from rising treatment costs and growing chronic disease prevalence. This project builds machine learning models to predict patients' annual healthcare expenditures, helping: - **Insurance companies** estimate risk and set fair premiums - **Hospitals** allocate resources more efficiently - **Policymakers** identify and address key drivers of healthcare spending The best model — **Gradient Boosting** — achieves **R² ≈ 0.887**, an RMSE of ~2,538 EGP, and an MAE of ~1,995 EGP on held-out test data. --- ## Project Structure ``` egypt-healthcare-cost-prediction/ │ ├── Egypt_healthcare_cost_prediction_enhanced.ipynb # Main notebook ├── egypt_healthcare_synthetic_data.csv # Dataset │ ├── eda_overview.png # EDA grid ├── model_comparison.png # Model comparison chart ├── prediction_diagnostics.png # Actual vs Predicted & residuals ├── feature_importance.png # Feature importance plot │ ├── requirements.txt └── README.md ``` --- ## Dataset > **This dataset is synthetically generated for educational purposes.** It does not represent real patient records. | Feature | Type | Description | |---|---|---| | `age` | Numeric | Patient age | | `bmi` | Numeric | Body mass index | | `num_dependents` | Numeric | Number of dependents | | `chronic_conditions` | Numeric | Number of chronic diseases | | `visits_per_year` | Numeric | Annual healthcare visits | | `smoker` | Binary | Smoking status | | `diabetes` | Binary | Diabetes diagnosis | | `hypertension` | Binary | Hypertension diagnosis | | `obese` | Binary | Obesity status | | `insured` | Binary | Insurance coverage | | `sex` | Categorical | Patient sex | | `region` | Categorical | …