From Student Teachers to Hired Educators: Exploring the Attributes of Successful Teacher Certification and Placement Through Data Mining

Periodic checkpoints for teacher quality are necessary.We collected metadata from 2551 students registered in a teacher education program in Taiwan. Rules for predicting success at certification and placement checkpoints were generated based on learning algorithms found to have the best model performances. The results from the decision trees showed that across two checkpoints, student teachers’ pedagogical content knowledge served as a successful rule for student groups from different academic schools. Other successful rules showed that teachers who found placements had comparatively higher performance scores in courses in which they would have traditionally been weak, based on their discipline. The methods used in this study, including metadata pre-processing and decision trees, can be applied to future studies in teaching and learning.