Comparative Models of Price Estimation Using Multiple Linear Regression and Random Forest Methods

Abstract

The house is one of the essential humans needs as a place to gather and do activities with family, and shelter, as well as a means of investment. The growth rate of people's demand for housing, especially houses in an area, is influenced by the rate of population growth in that area. Some regions in Indonesia with a reasonably high population growth rate are Jakarta, Bogor, Depok, Tangerang, and Bekasi (Jabodetabek). On the other hand, property entrepreneurs must be able to project house prices because businesses engaged in the property sector are currently very competitive. This study aims to model and compare several machine learning methods to estimate house prices in Jabodetabek based on facilities, year of construction, location, land and building area, number of rooms, condition of house construction, and legality documents. This modeling uses Multiple Linear Regression and Random Forest methods. The results of the modeling evaluation where the Random Forest model has an accuracy rate of 95.6%, while the Multiple Linear Regression model has an accuracy rate of 75%.