Determination of Residential Type in Subdivision Projects by the Decision Tree Analysis

Abstract

At present, to determine the residential type in a housing development project requires the experience and expertise of the project developers. This research aims to apply the Decision Tree Analysis to help analyze the residential type that will make the subdivision project achieve a good sales rate. The research process was conducted by collecting data from 179 residential subdivision projects from the market reports of real estate companies in Thailand, and then selected those projects with higher average monthly sales rates than those of all projects. A total of 59 projects were selected, comprising 31 townhouse, 22 detached house and 6 semi-detached house projects. Data for each project were collected. Then, these data sets and factors were analyzed using the Decision Tree Analysis. The results showed that the factors used in determining the residential type were the distance from the bus stations, land appraisal value and distance from parks. The model has an accuracy of 81.82% and is useful for those involved in the real estate development industry because it logically shows the decision- making process. It also demonstrates the potential for applying machine learning technique, as a technological innovation, to create models that aid real estate development decisions related to the customer’s behavior.