RT Journal Article
T1 Real-Time Prediction of Gamers Behavior Using Variable Order Markov and Big Data Technology: A Case of Study
A1 Baldominos Gómez, Alejandro
A1 Albacete García, Esperanza
A1 Marrero, Ignacio
A1 Sáez Achaerandio, Yago
AB This paper presents the results and conclusions found when predicting the behavior of gamers in commercial videogames datasets. In particular, it uses Variable-Order Markov (VOM) to build a probabilistic model that is able to use the historic behavior of gamers and to infer what will be their next actions. Being able to predict with accuracy the next user's actions can be of special interest to learn from the behavior of gamers, to make them more engaged and to reduce churn rate. In order to support a big volume and velocity of data, the system is built on top of the Hadoop ecosystem, using HBase for real-time processing; and the prediction tool is provided as a service (SaaS) and accessible through a RESTful API. The prediction system is evaluated using a case of study with two commercial videogames, attaining promising results with high prediction accuracies.
PB IMAI Software Research Group
SN 1989-1660
YR 2016
FD 2016-03
LK https://hdl.handle.net/10016/22491
UL https://hdl.handle.net/10016/22491
LA eng
NO This work is part of Memento Data Analysis project, co-funded by the Spanish Ministry of Industry, Energy and Tourism with identifier TSI-020601-2012-99 and is supported by the Spanish Ministry of Education, Culture and Sport through FPU fellowship with identifierFPU13/03917
DS e-Archivo
RD 24 may. 2024