🏠 Real Estate Price Prediction using ML**

This is a full-stack machine learning project to predict real estate prices based on user input. The application allows users to input property details via a simple front-end and returns predicted prices using a trained ML model served by a Flask backend.

📁 Project Structure

RealEstateMLProject/ │ ├── Client/ # Front-end interface (HTML/CSS/JS or framework) ├── Server/ # Flask API backend ├── Model/ # Jupyter Notebooks, dataset, columns and trained model pickle file ├── BHP.csv # Dataset used └── README.md # Project overview

Workflow

1. User Input
   The user enters details like location, BHK, sqft, etc. via the front-end interface.

2. API Request
   The data is sent to a Flask backend via an HTTP request.

3. Model Prediction
   The server loads a trained regression model and returns the predicted property price.

4. Result Display
   The prediction is shown to the user in a user-friendly format.

🛠 Technologies Used

• Python
• Flask(back-end)
• Pandas, NumPy, Scikit-learn
• HTML/CSS/JS (basic front-end)
• Git & GitHub

📊 Dataset

-Used a tabular dataset from kaggle

• Link: https://www.kaggle.com/amitabhajoy/bengaluru-house-price-data
• Source: BHP.csv
• Contains real estate data from Bangalore including various raw unprocessed features and price.

🧠 Model Information

• Model Used: The optimal model after fine-tune on GridSearchCV on few of standard ML regressor models like Lasso, DecisionTree, etc and grid of parameters.
• Libraries: pandas, numpy, scikit-learn, matplotlib, seaborn
• Feature Engineering: Added price per sqft which is important feature as per domain knowledge
• Preprocessing: Handling null values, cleaning and formatting columns having raw data
• Outlier Removal:Using domain knowledge and k-standard deviation techniques -Dimensionality Reduction: Labelling all the locations having very few records as others and then performing One-Hot Encoding -Training and Validation:Used R2 score and k-fold cross validation -Testing and Prediction:Define a custom function for prediction and testing -Saving:Model is saved as pickle file

How to Run

1. Clone the repo:

   Execute this command to clone the repo

   git clone https://github.com/avkhalkar/RealEstateMLProject.git

Set up a virtual environment and install dependencies, provided in requirements.txt and pythonversion.txt.

Run the Flask server:

cd Server python app.py Open Client/index.html in a browser to access the UI.

Screenshot of the web app

License

This project is licensed under the MIT License. You are free to use, modify, and distribute this software with proper attribution.