Karachi House Price Prediction ML App

This project is a machine learning application that predicts house prices in Karachi based on user inputs such as location, number of bedrooms, bathrooms, square footage, and more. The model is built using a Random Forest Regressor and is deployed using Streamlit.

Live Demo

Check out the live app here!

Table of Contents

• Overview
• Features
• How to Use
• Steps Performed
• Model Training
• Technologies Used

Overview

The Karachi House Price Prediction App allows users to input key details of a property, including location, number of rooms, and square footage, and instantly receive a predicted house price. This tool aims to provide an estimate based on historical data, making it easier for users to make informed decisions about real estate in Karachi.

Features

• Location Selection: Choose from various popular locations in Karachi.
• House Details Input: Enter the number of bedrooms, bathrooms, and the size of the house in square yards.
• Prediction: Get an instant price estimate based on the Random Forest Regressor model.
• Purpose and Property Type: Choose the purpose of the property (For Sale/For Rent) and the type (House, Flat, etc.).
• Interactive User Interface: A modern and responsive interface built with Streamlit.

How to Use

1. Navigate to the app using the live link or run it locally.
2. Input the required property details:
   • Select the location of the house.
   • Input the number of bedrooms, bathrooms, and square yards.
   • Select the property purpose (For Sale/For Rent).
   • Choose the property type (House, Flat, etc.).
3. Click on the "Predict House Price" button to get an estimate.
4. The predicted price will be displayed in the app.

Steps Performed

1. Data Preprocessing:

   • Handled missing and duplicate values.
   • Remove outliers and abnormal data points.

2. Feature Engineering:

   • Converted and scaled continous features e.g prices, sq_feets, sq_yards.
   • Applied one-hot and ordinal encoding to categorical variables for better model interpretation.

3. Model Building:

   • Selected Random Forest Regressor as the best predictive model among Linear Regression models and XGBoost Regressor.
   • Split the data into training and test sets.
   • Trained the model using the training dataset.

4. Hyperparameter Tuning:

   • Performed Grid Search with K-Fold Cross-Validation to optimize model parameters.
   • Tuned key hyperparameters such as the number of trees, maximum depth etc.

5. Model Evaluation:

   • Evaluated the model using metrics such as R² score, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE).
   • Selected the best-performing model for deployment.

6. Application Development:

   • Built a Streamlit app with an interactive UI to accept user inputs.
   • Integrated the trained Random Forest model to generate predictions based on user inputs.

7. Deployment:

   • Deployed the app on Streamlit, making it accessible through the web.

Model Training

The Random Forest model was trained on historical house price data, with the following features:

• Location
• Number of bedrooms
• Number of bathrooms
• Size of the house (square yards)
• Property type and purpose

For hyperparameter tuning and model evaluation, a Grid Search with K-Fold Cross-Validation was applied to ensure optimal performance.

Technologies Used

• Python: Backend programming language.
• Streamlit: For building the web interface and deploying the app.
• scikit-learn: For training the machine learning model (Random Forest Regressor).
• Pandas & NumPy: For data manipulation and preprocessing.
• Matplotlib & Seaborn: For data visualizations.
• Git: For version control.