Toward Personalized Adaptive Gamification: A Machine Learning Model for Predicting Performance

Abstract

Personalized adaptive gamification has the potential to improve individuals' motivation and performance. Current methods aim to predict the perceived affective state (i.e., emotion) of an individual in order to improve their motivation and performance by tailoring an application. However, existing methods may struggle to predict the state of an individual that it has not been trained for. Moreover, the affective state that correlates to good performance may vary based on individuals and task characteristics. Given these limitations, this paper presents a machine learning method that uses task information and an individual's facial expression data to predict his/her performance on a gamified task. The training data used to generate the adaptive-individual-task model is updated every time new data from an individual is acquired. This approach helps to improve the model's prediction accuracy and account for variations in facial expressions across individuals. A case study is presented that demonstrates the feasibility and performance of the model. The results indicate that the model is able to predict the performance of individuals, before completing a task, with an accuracy of 0.768. The findings support the use of adaptive models that dynamically update their training data set and consider task information and individuals' facial expression data.