Spatial Regression Model Predicting Condominium Price per Square Meter in Bangkok

Abstract

Setting the condominium’s selling price per square meter before project launching is a big challenge for real estate developers. This current study aims at developing the spatial regression model predicting the average condominium’s selling price per square meter in Bangkok in 2015 from 652 projects. Three models were tested in this research; namely Ordinary Least Square (OLS), spatial lag, and spatial error regression model to find the best-fitting model. There are 17 predictors such as location along the Chao Phraya river bank, the distance to the BTS station, the number of road traffic lanes in front of project’s location and the numbers of condominium floors, but geographical coordinate used as spatial weight is excluded. Both spatial regression models (R 2 = 0.72 & 0.73 for spatial lag and spatial error respectively), fit data better than the OLS model. Seven variables significantly and positively correlate with average price such as location on the east side of the Chao Phraya River, numbers of condominium floors, condominium project area, furniture installation, condominium project launch in 2015 and development by public company. Two variables significantly and negatively correlate with average price such as a total number of units, the distance to the BTS station and the distance to the MRT station.