Housing Price Prediction: Regression Analysis & Model Comparison

Project Overview

This repository analyzes residential property sales data (2006-2008) to identify key determinants of housing prices. Through exploratory data analysis and regression modeling, we:

1. Evaluate different functional forms for price prediction
2. Compare time-based vs. structural feature models
3. Develop a high-performance prediction model (R² = 0.828)

Key Insights:

• Structural features (quality, size) dominate temporal effects
• Log-log specifications best capture non-linear relationships
• Year fixed effects have minimal predictive power

Data Inputs

• Primary dataset: housing_train.csv (1,941 sales records)
• 80+ features including:
  • Structural: Lot area, living space, quality ratings
  • Temporal: Year built, year sold
  • Location: Neighborhood, zoning
• Target variable: v_SalePrice ($13,100 - $755,000 range)

Key Visualizations

Top predictors of sale price

Price premium by quality rating

Installation & Requirements *

pip install pandas
pip install numpy
pip install statsmodels
pip install matplotlib
pip install seaborn

*Make sure to download the summary_colFE.py file