Analysis of the Pedagogical Effectiveness of Teacher Qualification Cycle in Morocco: A Machine Learning Model Approach

Abstract

Given the importance of education, and their related professions such as teaching qualification. We attempt in this paper to investigate the pedagogical effectiveness of trainee teachers of primary and secondary grades. ICT use, teaching level, gender and score of graduation are used as features, while category of effectiveness is used as the predicted variable. To achieve this goal, we introduce machine learning techniques, which enable the development and use of algorithms and statistical models to analyze and infer pattern in data. In this study, we introduce K-Nearest Neighbor, Multi-layer Perceptron Neural Network, and Gradient Boosting as supervised learning and K-means clustering as unsupervised learning algorithm to show the performance of these models in predicting the pedagogical effectiveness of future Moroccan teachers. As results, we found that gradient boosting classification perform very well than the other algorithms. Moreover, it can predict teacher effectiveness based on all above features, evaluation of the models’ performance was performed according to seven criteria (Precision score, recall score, F1 score, accuracy score, macro average, micro average, and weighted average), which confirm the goodness of gradient boosting classifier. It is noteworthy that changing the parameters of models can provide different results, which can be subject of further works.