The research group of Professor Yasumasa Matsuda (Graduate School of Economics) and Assistant Professor Uki Sato (Yotta Informatics Research Center) of Tohoku University and Propre Japan Co., Ltd. have started joint research on the application of spatiotemporal models in the real estate leasing market.Utilizing the knowledge about the spatio-temporal model owned by Matsuda-Sato Laboratory and the big data owned by Propre, we analyze the determinants of the solicitation period and rent unit price in rental real estate.Create a unique algorithm for real estate valuation.
The real estate leasing market is greatly affected by the times, and it is thought that the "appropriateness" of prices will change between February-March recruitment, when demand increases due to enrollment and transfer, and April and after, when demand has subsided. ..In addition, the influence from the geographical environment is also large, for example, if a large number of similar rental solicitations are made in the surrounding area, there is a possibility that the risk of a long period until the contract is concluded will increase for the same rent.In general statistical / machine learning models, it is assumed that each data is temporally and spatially independent, but this assumption does not hold for real estate data.
Therefore, in real estate data analysis, introduction of a spatiotemporal model that can explicitly incorporate time-space-dependent correlations and empirical research are required.By realizing this, it becomes possible to analyze the real estate at a certain point in time series, taking into account changes in the surrounding environment.In other words, it can lead to the development of services such as rent assessment and real estate valuation with higher accuracy.
Through this joint research, Propre will research and develop more appropriate assessment algorithms for various stakeholders, with a view to publishing research results in papers in the future, and use it for the development of new services in the future. I plan to go.In addition, we will develop services that are useful to people in various ways using the real estate database, and will continue to carry out research activities.
Professor Matsuda of Tohoku University has published many papers after many years of research on spatiotemporal models, and in 2019, he received the 13th Research Achievement Award of the Japan Statistical Society for his research on applied analysis of spatiotemporal models. Acquired.Currently, he is the director of the Service and Data Science Research Center established by Tohoku University to promote industry-academia joint research.
Propre collects and organizes more than 17 million real estate information per day from 1 countries around the world, and is one of the largest in the world in terms of the number of real estate information handled.By aggregating and analyzing data collected from each country from a unique perspective, we will develop a platform that anyone can easily "reach to real estate data in the world", "buyers, sellers, borrowers, lenders". proceeding.
