Cook County Housing Price Prediction

EDA, feature engineering, and linear regression on 200K+ housing records from Cook County, Illinois

Overview

This two-part project explores what can be learned from an extensive housing dataset embedded in a dense social context in Cook County, Illinois (the county where Chicago is located). The dataset contains over 500,000 records and 62 features, with the goal of predicting residential sale prices using linear regression.

Part 1: Exploratory Data Analysis & Feature Engineering

  • Investigated the granularity, provenance, and social context of the Cook County Assessor’s Office (CCAO) dataset
  • Performed EDA on 204,792 training observations using pandas, matplotlib, and seaborn
  • Engineered new features and designed a custom error metric to replace RMSE for this setting
  • Analyzed the implications of adding demographic features (race, income, occupation) on model fairness

Key insight: Proposed proximity to subway stations as a domain-specific feature for predicting housing prices in dense urban environments, drawing on personal context growing up in Shanghai.

Part 2: Modeling & Analysis

  • Built and fit a linear regression pipeline using scikit-learn on cleaned and engineered features
  • Evaluated model error and analyzed residuals to identify systematic biases
  • Considered the real-world implications of predictive modeling in the context of Cook County’s history of racial discrimination in property taxation

Tools & Skills

Python · pandas · NumPy · scikit-learn · seaborn · matplotlib · feature engineering · linear regression · EDA