[4A2] [General Session] 2. Machine Learning

[4A2-03] An assessment of real estate price using neural network on REINS

〇Hikaru Fukui¹, Kazuyoshi Sakai¹, Tadataka Minamimura¹, Jyunichi Mio¹, Akihiro Kinoshita¹, Shiro Takata² (1. KINKI Real Estate Information Network System, 2. KINDAI University Faculty of Science and Engineering)

Keywords:The assessment of real estate price, data normalization, neural network, error back propagation of neural network

The purpose of REINS (Real Estate Information Network System prescribed by Ministry of Land, Infrastructure and Transport)is to distribute real estate and to contribute to improve the house environment of the people so that develop the market utilizing the analysis based on the data of real estate and the agreement real estate. The goal of this paper is to propose the machine learning method for an assessment of real estate price using neural network with agreement real estate on REINS. First, this paper describes the classification and normalization of input data after current state of real estate price assessment. Data normalization for neural network is required to be numeric data from 0 to 1 because most input items are text data. After analyzing the experiment data, we try to realize how the data affect real estate price. Then, we perform error back propagation of neural network with agreement real estate as input data, and classify per million, half a million, 100 thousand. Finally, the experiment enhances feasibility.