🏠 House Price Prediction System

A complete end-to-end machine learning project for predicting house prices using Linear Regression on the real-world Kaggle Ames Housing Dataset, complete with a beautiful Streamlit web interface for deployment.

📋 Table of Contents

• Overview
• Features
• Project Structure
• Installation
• Usage
• Model Details
• Results
• Technologies Used

🎯 Overview

This project demonstrates a professional machine learning pipeline. It predicts house prices based on real-world features like above-ground living area, bedrooms, bathrooms, year built, overall material quality, and specific neighborhoods in Ames, Iowa.

✨ Features

• Real-World Data: Utilizes the industry-standard Kaggle Ames Housing Dataset.

• Robust Preprocessing Pipeline: Implements scikit-learn Pipelines with SimpleImputer, StandardScaler, and OneHotEncoder to handle missing values and categorical data automatically.

• Model Training: Linear Regression with strict training/testing splits to prevent data leakage.

• Visualizations:

• Correlation matrix

• Feature distributions

• Prediction plots

• Feature importance

• Interactive Web App: Streamlit interface with dynamic dropdowns and real-time predictions.

• Model Persistence: Save and load the entire trained pipeline using Joblib.

• Performance Metrics: R², RMSE, MAE, MSE.

📁 Project Structure

house_price_prediction/
│
├── data/                          # Data directory
│   └── train.csv                 # Kaggle Ames Housing Dataset
│
├── models/                        # Saved models
│   ├── house_price_model.pkl     # Trained ML Pipeline (Imputer + Scaler + Encoder + Model)
│   └── metrics.json              # Model metrics
│
├── notebooks/                     # Visualizations
│   ├── correlation_matrix.png
│   ├── feature_distributions.png
│   ├── predictions_plot.png
│   └── feature_importance.png
│
├── src/                           # Source code
│   ├── preprocessing.py          # Data preprocessing components
│   ├── model.py                  # Model training & evaluation
│   └── visualization.py          # Plotting functions
│
├── train_model.py                # Main training pipeline script
├── app.py                        # Streamlit web application
├── requirements.txt              # Python dependencies
└── README.md                     # This file

🚀 Installation

Prerequisites

• Python 3.8 or higher
• pip package manager

Step 1: Clone or Download the Project

git clone https://github.com/Imhs14/House-Price-Prediction.git
cd House-Price-Prediction

Step 2: Install Dependencies

pip install -r requirements.txt

Step 3: Download the Dataset

Download train.csv from the Kaggle House Prices Competition and place it inside the data/ folder.

💻 Usage

1. Train the Model

python train_model.py

This script will:

• Load the Kaggle dataset.
• Build and fit the preprocessing pipeline (imputation, scaling, encoding).
• Train the Linear Regression model.
• Evaluate performance on unseen test data.
• Save the entire pipeline and metrics.

2. Run the Streamlit App

streamlit run app.py

The web application will open in your browser (usually at http://localhost:8501).

🤖 Model Details

Algorithm

Linear Regression - Mapped through a scikit-learn Pipeline to seamlessly handle raw input data directly from the user interface.

Input Features

1. GrLivArea - Above grade (ground) living area square feet
2. BedroomAbvGr - Number of bedrooms above basement level
3. FullBath - Full bathrooms above grade
4. YearBuilt - Original construction date
5. OverallQual - Overall material and finish quality (1-10)
6. Neighborhood - Physical locations within Ames city limits (Categorical)

Target Variable

SalePrice - Property sale price in USD

Preprocessing Pipeline

• Missing Numeric Data: Imputed using the median strategy.
• Missing Categorical Data: Imputed using the most frequent strategy.
• Numeric Scaling: StandardScaler for normalization.
• Categorical Encoding: OneHotEncoder to translate text-based neighborhoods into machine-readable formats.
• Train-Test Split: 80-20 split.

📊 Results

The model achieves realistic and strong performance on real-world housing data:

• R² Score: ~0.85 (explains 85% of variance in real-world prices)
• RMSE: ~$35,000
• MAE: ~$25,000

Note: These metrics reflect the natural noise and variance found in actual real estate markets, avoiding the overfitting commonly seen with synthetic datasets.

🛠️ Technologies Used

Machine Learning & Data Science

• Scikit-learn: Pipelines, Transformers, and Linear Regression
• Pandas: Data manipulation and ingestion
• NumPy: Numerical operations

Visualization

• Plotly: Interactive charts and gauges
• Matplotlib / Seaborn: Statistical visualizations

Web Application

• Streamlit: Web interface framework

Model Persistence

• Joblib: Pipeline serialization

📈 Key Insights

Based on real-world data analysis:

• Overall Quality and Living Area are the strongest predictors of sale price.
• Neighborhood plays a massive role in valuation, requiring categorical encoding to capture properly.
• Age (Year Built) shows a clear depreciation curve for older homes that haven't been remodeled.

🔮 Future Enhancements

• Add more advanced algorithms (Random Forest, XGBoost) to capture non-linear relationships.
• Expand the feature selection to include basement square footage and garage condition.
• Implement an automated hyperparameter tuning grid.
• Add data upload functionality for bulk predictions.

🤝 Contributing

Feel free to fork this project and submit pull requests for improvements!

📄 License

This project is open source and available for educational purposes.

---

Happy Predicting! 🏡💰*