rajyan/gharDAAM

---
license: mit
library_name: scikit-learn
pipeline_tag: tabular-regression
tags:
  - machine-learning
  - regression
  - scikit-learn
  - random-forest
  - house-price-prediction
  - real-estate
  - india
---

# 🏠 gharDAAM – House Price Prediction Model

## Overview

**gharDAAM** is a machine learning project that predicts residential property prices in India using structured real estate data. The model is built using **Random Forest Regression** and trained on a cleaned and preprocessed dataset containing approximately **175,000** property listings.

This repository contains the trained model and documentation required to reproduce predictions.

---

## Model Information

| Property | Value |
|----------|-------|
| Task | House Price Prediction |
| Problem Type | Regression |
| Algorithm | Random Forest Regressor |
| Framework | scikit-learn |
| Language | Python |
| Target Variable | Property Price |

---

## Model Performance

| Metric | Score |
|--------|------:|
| R² Score | **0.82** |
| Mean Absolute Error (MAE) | **₹17.1 Lakh** |

The Random Forest model was selected after evaluating multiple regression algorithms because it provided the best balance between predictive accuracy and generalization.

---

## Features Used

The model was trained using features such as:

- BHK
- Location
- Transaction Type
- Furnishing Status
- Bathroom Count
- Balcony Count
- Covered Parking
- Open Parking
- Garden/Park Availability
- Main Road Facing
- Swimming Pool Availability

---

## Requirements

Recommended package versions:

```text
Python >= 3.11

scikit-learn == 1.7.2
pandas
numpy

Using a different scikit-learn version may generate compatibility warnings when loading the serialized model.

---

Loading the Model

import pickle

with open("house_price_predictor_model.pkl", "rb") as f:
    model = pickle.load(f)

---

Example Prediction

import pandas as pd
import pickle

with open("house_price_predictor_model.pkl", "rb") as f:
    model = pickle.load(f)

sample = pd.DataFrame([{
    "BHK": 2,
    "Location": "thane",
    "Transaction": 0,
    "Furnishing": "Semi-Furnished",
    "Bathroom": 2,
    "Balcony": 1,
    "Covered Parking": 1,
    "Open Parking": 0,
    "Garden/Park": 1,
    "Main Road": 0,
    "Pool": 0
}])

prediction = model.predict(sample)

print(prediction)

---

Repository Contents

.
├── house_price_predictor_model.pkl
├── README.md

If preprocessing artifacts (such as encoders, scalers, or imputers) are required for inference, they should also be included in the repository.

---

Dataset

The model was trained using a cleaned and preprocessed version of a publicly available Indian residential real estate dataset.

The original dataset was obtained from Kaggle and licensed under the Community Data License Agreement – Sharing – Version 1.0 (CDLA-Sharing-1.0).

This repository does not redistribute the original dataset.

---

Limitations

• Predictions are based only on the available training features.
• The model should be used for educational and research purposes.
• Real estate prices are influenced by market conditions that may change over time.
• Performance may decrease for regions or property types underrepresented in the training data.

---

Future Improvements

• Feature importance analysis
• Model deployment using FastAPI
• Interactive web interface
• Continuous retraining with updated data

---

Acknowledgements

• The original real estate dataset was sourced from Kaggle uploaded by Juhi Bhojani.

• Thanks to the open-source Python ecosystem, especially:

  • scikit-learn
  • pandas
  • NumPy

---

License

This repository is released under the MIT License.

Please refer to the original dataset's license for any restrictions related to the training data.

---

Author

Raj Aryan

Electronics and Communication Engineering (ECE) Indian Institute of Information Technology (IIIT) Surat

Interested in:

• Artificial Intelligence
• Machine Learning
• Robotics
• Data Science