[4G3-OS-8b] マイニングと知識創発(2)

[4G3-OS-8b-02] Predictability of Number of Visitors from GPS Location Big Data

野球興業の来場者数予測モデルの構築を中心に

Hirotoshi Yanagi^1,3, 〇Yoshitoshi Nakahara², Takahiro Hoshino^2,3 (1. Keio University, Graduate school of Economics, 2. Keio University, Faculty of Economics, 3. Riken AIP)

Keywords:GIS, Location big data, Data Mining, Flow Population

It is the purpose of this paper to discuss whether we could predict the number of visitors of public facilities from the GPS location big data collected from approximately 20 million smartphones in total while taking into consideration influences of various covariates to correct several biases relate to the penetration ratio. We used number of visitors of baseball games in the year 2018 as a label, and GPS location big data and other features as feature vectors to apply supervised machine learning methods. Because of the prediction accuracy, we concluded that it is possible to predict number of visitors of public facilities from GPS location big data.