Empirical Valuation and Opportunity Detection in Emerging South Asian Real Estate: A Case Study of DHA Lahore Using Explainable AI (TreeSHAP)

Abstract

Automated Valuation Models (AVMs) are transforming property underwriting and investment benchmarking, yet their application to unorganized emerging real estate markets in South Asia remains critically under-researched. In planned urban developments such as the Defence Housing Authority (DHA) Lahore, plot transactions are characterized by steep micro-locational volatility across phases, blocks, and bespoke amenities that classical linear appraisals fail to capture. This paper presents an empirical machine learning benchmark evaluating Ordinary Least Squares (OLS) Multiple Linear Regression against tree-based ensembles (Random Forest, XGBoost, and LightGBM) on a curated dataset of residential plots across DHA Lahore Phases 2 through 9 Prism. Incorporating smoothed target-encoded block hierarchies alongside micro-locational features, Random Forest demonstrated superior out-of-sample performance, achieving a Test R² of 0.7499 and an RMSE of 2.0395 Crore PKR, outperforming linear baselines by over 51 percentage points.