Rental House Price Predictor Web App

This web app predicts rental house prices based on user-provided features such as the number of bedrooms, square footage, etc. It uses a RandomForestRegressor model and is built with Flask for the backend.

Features

• Predict rental house prices based on user input.
• Real-time predictions powered by a pre-trained Random Forest model.
• User-friendly interface with dropdowns and input fields.

Prerequisites

Prerequisites:

• Python 3.10 or above.
• Flask framework installed.
• Libraries: Scikit-learn, Pandas, Joblib and Gunicorn.
• A web browser for accessing the UI.

Installation Guide

1. Clone the repository

       git clone https://github.com/bhaskrr/rental_house_price_predictor.git
       cd rental_house_price_predictor

2. Install dependencies

       pip install -r requirements.txt

3. Run the app

       gunicorn app:app

4. Open the app in a web browser: Visit http://127.0.0.1:8000.

Usage

1. Open the app in a browser.
2. Fill in the required details:
   • Layout Type (select from dropdown).
   • Property Type (select from dropdown).
   • Furnish Type (select from dropdown).
   • Area in Square foot.
   • Number of bedrooms.
   • Number of bathrooms.
3. Click on the Get Rent button.
4. View the predicted rental price.

Model Information

• Algorithm: RandomForestRegressor.
• Training Data: Dataset with 193011 rental listings, including features like layout type, property type, bedrooms, square footage, and rental price etc.
• Dataset Link: https://www.kaggle.com/datasets/saisaathvik/house-rent-prices-of-metropolitan-cities-in-india/data?select=_All_Cities_Cleaned.csv.
• Evaluation: 0.77 r2 score on test data.

Project Structure

rental_house_price_predictor/
|
|- app.py                           # Main Flask App
|- requirements.txt                 # Dependencies
|- rf_regressor.zip                 # Compressed Pre-trained RandomForestRegressor Model
|- furnish_type_encoder.joblib      # Pre-fitted Furnish Type Encoder
|- layout_type_encoder.joblib       # Pre-fitted Layout Type Encoder
|- property_type_encoder.joblib     # Pre-fitted Property Type Encoder
|- static/                          # Static Files
|   |__ index.css
|- templates                        # HTML Templates
|   |__ index.html
|- README.md                        # Documentation

Known Issues and Limitations

1. Predictions are less accurate for out-of-distribution data (e.g., very large houses).

2. No support for categorical variables outside the trained dataset.