Philadelphia Housing Prediction Model Analysis [Collaboration]

Takeaway: Incorporating neighborhood and structural context improves housing price predictions, but model performance varies spatially and exhibits diminishing returns with added complexity.

---

Overview This project was completed as a team-based midterm for MUSA 5080: Public Policy Analytics and evaluates alternative strategies for predicting residential housing prices in Philadelphia. The analysis focuses on comparing model performance across different feature sets and examining how prediction accuracy varies spatially across the city.

---

Key Outputs

• Analytical report:
• Quarto slides
• Code repository:

---

Key Findings

• Models incorporating neighborhood context outperform property-only baselines.
• Gains from additional model complexity are uneven and diminish beyond core predictors.
• Prediction errors vary spatially, indicating differential reliability across housing submarkets.