Main Variables in Teaching Quality: A Data Mining Approach (#310)

Authors

Castrillon, Omar Danilo

Abstract

Abstract – This research is based on a student perception survey that was applied to students about their university professors. This survey consists of 1 dependent variable (teaching quality) and 20 independent variables: time dedicated to the subject, reasons for enrollment, adequate class preparation, adequate mentorship, fostering of relationships with other subjects, pleasant work methods, encouragement of argumentation, promotion of autonomous learning, motivation of the topics covered, respectfulness, attention that everyone learned, adaption of teaching methods, respect of the rules, fairness, students learnt with evaluations, depth of learning, the professor transformed their way of thinking in regards to the course, the professor taught the students another course, learning was facilitated by the professor’s strengths, the professor endeavored for students to learn. Through the J48 algorithm of the free machine learning and data mining platform called Weka, the main variables are identified for each group of students that are most important in this survey. As a result of this research, three groups of students are found, with three different sets of independent variables capable of predicting the teaching quality. In the first group of students, with an effectiveness greater than 96%, the following variables are found: adequate class preparation, adaptation of teaching methods, reasons for enrollment, fairness, learning was facilitated by the professor’s strengths, depth of learning, respectfulness. In the second group of students, with an effectiveness greater than 84%, the following variables are found: learning was facilitated by the professor’s strengths, depth of learning, adequate class preparation. Finally, in the third group, with an effectiveness greater than 78%, the following variables are established: learning was facilitated by the professor’s strengths and adequate mentorship.