Presentation information

General Session

General Session » [GS] J-2 Machine learning

[1J4-J-3] Data mining: social mining

Tue. Jun 4, 2019 5:20 PM - 6:20 PM Room J (201B Medium meeting room)

Chair:Shogo Okada Reviewer:Tomoya Yoshikawa

5:20 PM - 5:40 PM

[1J4-J-3-01] Modelling of retention of media service users based on power-law

〇Soichiro Morishita1 (1. CyberAgent, Inc.)

Keywords:statistical modeling, retention rate, power law

In this paper, a model of the visitor retention in media services via the Internet such as video distribution service is shown. Visitor intervals of media services usually follow power-law. Based on the theory, we describe the modeling of visitor intervals and separate the residual into the elements of return and retention. Moreover, the result of the separation according to access log of the Internet TV station "AbemaTV" is shown as an example.