Redfin data of SoCal

This dataset supports the research article “Predicting Residential Property Values Using XGBoost and Spatial–Temporal Encoding: Evidence from Southern California” by Michael Quintana (2025).The dataset contains a cleaned and anonymized subset of residential property transactions derived from Redfin’s publicly available data export (June 2025).Each observation represents a single-family home, condominium, or townhouse sold in Southern California.Variables include sale price, living area, lot size, year built, bedrooms, bathrooms, ZIP code, and days on market.The dataset was used to train and validate an XGBoost regression model designed to estimate home prices using both structural and spatial–temporal features.All personally identifiable or proprietary location data have been removed or aggregated at the ZIP-code level to maintain privacy while preserving statistical utility.This dataset and accompanying R scripts allow replication of the core results presented in the study, including model training, feature importance analysis, and predictive performance evaluation.