Housing Market Prediction

Predicting Demand and Supply of Housing Markets in the United States using PySpark and Machine Learning.

Project Overview

This project aims to predict housing market conditions (buyers' market, sellers' market, or neutral market) using machine learning techniques and PySpark. We analyze multiple factors including:

• Housing market metrics from Zillow
• Building permits data
• Treasury bill rates
• Bank failure information

Data Sources

• Zillow Research Data: https://www.zillow.com/research/data/
• Building Permits: https://www.census.gov/construction/bps/historical/state_units.html
• Treasury Bill Rates: https://home.treasury.gov/resource-center/data-chart-center/interest-rates/
• Bank Failures: https://catalog.data.gov/dataset/fdic-failed-bank-list

Project Structure

housing-market-prediction/
├── data/               # Data files
├── notebooks/         # Jupyter notebooks
├── src/               # Source code
├── docs/             # Documentation
└── requirements.txt  # Project dependencies

Key Features

• Multi-source data integration using PySpark
• Machine learning models:
  • Logistic Regression
  • Random Forest
  • Naive Bayes
• Handling class imbalance
• Seasonal trend analysis
• Geographic market variation analysis

Results

• Achieved over 80% accuracy in market condition prediction
• Logistic Regression performed best with:
  • 87% precision for sellers' market
  • 85% precision for buyers' market
• Successfully identified seasonal patterns in market behavior

Setup and Installation

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/housing-market-prediction.git
cd housing-market-prediction

2. Install dependencies

pip install -r requirements.txt

Contributors

• Aadit Malikayil
• Bhavika Karale
• Kapil Tare
• Shreyas Kashyap