머신 러닝 방법을 이용한 오피스 임대료 산정 -랜덤 포레스트, 인공 신경망, 서포트 벡터 머신 활용을 중심으로-

초록

We estimate office market rent using random forests, artificial neural networks, and support vector machines, which are commonly used in recent studies related to automatic evaluation models. We examine each method and compare the performances based on our sample data of the 507 offices located in Seoul, and depict PD(Partial Dependence) Plots to identify the influence of variables on predicted values. Using random forests, artificial neural networks, and support vector machines method, we estimate office market rent determinant model based on 507 office rental information data in Seoul. We attempt to identify the best performed model and also adopt the Partial Dependence(PD) Plots to examine the relative impact of research variables on predicted value. We classify the sample data into Class A, Class B, Class C, and the below Class C group. We apply the same method for each of those group. In general, we find that the support vector machines model performs best followed by random forests model and artificial neural networks model. The results indicate that the rent has a positive relation with rentable/usable ratio, floor, building age, and number of elevator. The number of parking vehicles showed the quadratic curve on rent in the model. Our study contributes to compare and allows appraisers to have a cross validation process with traditional valuation model against the finding from a machine learning method on office market.

키워드