Exploration of Predictors for Teaching and Learning International Survey Teachers' Team Innovativeness Using glmmLasso, Machine Learning for Multilevel Data

• In Event: Data Science in Education

Abstract

As a newly added variable to TALIS 2018, teachers’ team innovativeness lacks empirical research. Exploratory research with machine learning is less likely to be bounded by theories or previous research, and particularly machine learning coupled with large-scale data can make contributions to the existing body of literature. To explore important predictors for teachers’ team effectiveness, we employed penalized regression, specifically glmmLasso, a machine learning technique to consider the multilevel data structure of TALIS in variable selection. In comparison to group LASSO and group Enet, glmmLasso showed stable results in terms of prediction and variable selection. Of 573 teacher and principal variables, glmmLasso selected 14 important predictors for teachers’ team innovativeness. Implications and future research topics are discussed.

Authors

• Miryeong Koo, Korea National University of Education

• Jin Eun Yoo, Korea National University of Education