Regression Machine Learning Projects

Welcome to my portfolio of Regression-focused machine learning projects. This repository serves as a central hub for projects where I explore real-world datasets, perform detailed exploratory data analysis (EDA), preprocess data, implement various regression models, and evaluate predictive performance.

This collection reflects a learn-while-doing approach, emphasizing hands-on application of regression techniques, interpretation of model results, and improvement through hyperparameter tuning. Each project is self-contained and showcases a complete workflow from data exploration to model evaluation.

---

Key Features & Insights (Summary)

Project	Key Features	Most Important Predictors	Best Performing Models	Key Insight
Medical Insurance Forecast	Age, Sex, BMI, Children, Smoker, Region	Smoking, Age, BMI	Gradient Boosting, Random Forest	Ensemble methods handle non-linear relationships well; smoking status is a major cost driver.
House Prices Model	LotArea, OverallQual, YearBuilt, Neighborhood, GarageCars, TotalBsmtSF	OverallQual, GrLivArea, YearBuilt	Gradient Boosting, Random Forest	Tree-based models outperform linear regression; feature importance reveals property quality drives prices.

This table allows quick understanding of each project's scope, predictive factors, and outcomes, ideal for a research portfolio.

---

Repository Overview

The repository currently includes two major projects:

Project Directory	Objective	Status
Medical Insurance Forecast/	Predicting medical insurance charges based on demographic and health features	Complete
housePricesModel/	Predicting house sale prices using property and location features	Complete

Each project folder includes:

• Jupyter Notebooks: Documenting the full machine learning workflow from EDA to model evaluation
• Model Files: Saved models, pipelines, and scalers (.joblib or .pkl)
• Data Files: Dataset used for analysis

---

📂 Projects Breakdown

1. Medical Insurance Forecast

Objective: Predict insurance charges based on age, BMI, smoking status, and other factors.

Workflow:

1. Exploratory Data Analysis (EDA): Visualizations, correlations, statistical tests (T-test, ANOVA).
2. Preprocessing: Outlier removal, scaling, and categorical encoding.
3. Modeling: Compared multiple regression techniques including Linear Regression, Ridge, Lasso, Huber Regression, Random Forest, and Gradient Boosting.
4. Feature Importance: Smoking, age, and BMI identified as the most influential features.
5. Evaluation: Model performance assessed via RMSE, R², residual plots, and learning curves.

Key Insight: Ensemble methods (Random Forest & Gradient Boosting) outperformed linear models, demonstrating robustness in predicting insurance costs.

---

2. House Prices Model

Objective: Predict house sale prices using property and neighborhood attributes.

Workflow:

1. EDA: Data visualization, feature correlations, handling missing values.
2. Preprocessing: Pipelines built with ColumnTransformer for combined encoding, imputation, and scaling.
3. Model Comparison: Linear Regression, Ridge, Lasso, Random Forest, and Gradient Boosting models evaluated.
4. Evaluation Metrics: Mean Squared Error (MSE), Mean Absolute Error (MAE), R² Score, Residual Plots, Actual vs Predicted plots.
5. Hyperparameter Tuning: Optimization using RandomizedSearchCV for best model performance.
6. Feature Interpretation: Feature importance analyzed for both linear and tree-based models.

Key Insight: Ensemble tree-based models captured complex non-linear relationships, outperforming linear models for house price prediction.

---

Technologies & Libraries

• Programming Language: Python
• Libraries: pandas, numpy, scikit-learn, matplotlib, seaborn, joblib, os
• Development Environment: Jupyter Notebook

---

📁 Repository Structure

Regression/
│
├── Medical Insurance Forecast/
│   ├── EDA.ipynb                        # Exploratory Data Analysis
│   ├── Medical Insurance Forecast Model.ipynb  # Modeling & evaluation
│   
│
├── housePricesModel/
│   ├── EDA.ipynb                        # Exploratory Data Analysis
│   ├── HousePriceModel.ipynb            # Modeling & evaluation
│  
│
└── README.md                             # Project documentation

---

Learning & Research Focus

This repository demonstrates:

• Hands-on application of regression techniques for real-world problems.
• Data-driven decision making: Statistical tests, feature importance, and correlation analysis.
• Model evaluation & optimization: Selecting best-performing models using metrics and hyperparameter tuning.
• Research mindset: Documenting workflow, interpreting results, and drawing actionable insights.

This approach mirrors student-led research projects, emphasizing both practical implementation and analytical understanding of predictive modeling.

---

How to Use

1. Clone the repository:

git clone <repository-url>

2. Open Jupyter notebooks and run sequentially:

   • EDA.ipynb → Explore data and insights
   • Model.ipynb → Preprocessing, modeling, evaluation, and tuning

---