Analysis of Coalition Formation in Cooperative Games Using Crowdsourcing and Machine Learning

Analysis of coalition formation in cooperative games is an important research topic in game theory. Previous studies on coalition formation used laboratory experiments to collect data on player decision making, but the amount of data collected was limited due to the high cost of laboratory experiments. In this study, we used crowdsourcing to collect a large volume of decision-making data for use in predicting player behavior in cooperative games. This large amount of data enabled us to train large machine learning models such as deep neural networks, which can more precisely predict player decision making in cooperative games. The results with our machine learning models using crowdsourced data were similar to those of laboratory experiments.