Classifying Preservice Teachers' Computational Thinking Skills Using Machine Learning Algorithms

• In Event: Models for Digital Learning in Computational and Game-Based Contexts

Abstract

Computational thinking (CT) skills of pre-service teachers have been explored extensively, but few methods have involved machine learning techniques to identify the relationships between CT predictors and CT skills. This study aims to compare and contrast the predictive capacity of four machine learning algorithms in classifying the CT skills of pre-service teachers. First, the results show that Decision Tree outperformed K-Nearest Neighbors, Logistic Regression, and Naive Bayes in predicting pre-service teachers’ CT skills. Second, the participants’ prior CT skills, time spent on CT training, and perceptions of difficulty regarding the learning content are the top three important predictors in this model. Identifying the most important factors in the preparation of future teachers could promote the development of CT skills.

Authors

• Hao-Yue Jin, University of Alberta
  Presenting Author

• Maria Cutumisu, McGill University
  Presenting Author