Predicting the Culturally Responsive Teacher Roles With Cultural Intelligence and Self-Efficacy Using Machine Learning Classification Algorithms

Abstract

This study aimed to predict the culturally responsive teacher roles based on cultural intelligence and self-efficacy using machine learning classification algorithms. The research group consists of 415 teachers from different branches. The Bayes classifier (NaiveBayes), logistic-regression (SMO), lazy-classifier (KStar), meta-classifier (LogitBoost), rule-learner (JRip), and decision-tree (J48) were employed in the assessment of the predictive model. The results indicated that JRip rule-learner had a better performance than other classifiers in predicting the culturally responsive teachers based on six attributes used in the study. The JRip rule-learner classified the culturally responsive teachers as low, medium, or high with an accuracy of 99.76% (CCI: 414/415) [Kappa statistic: 0.996, Mean Absolute Error (MAE): 0.003, Root Mean Square Error (RMSE): 0.043, Relative Absolute Error (RAE): 0.663, Relative Squared Error (RRSE): 9.244]. The results indicated that all classifiers had an acceptable performance but JRip rule-learner had a better performance than the other classifiers in predicting the culturally responsive teachers.