Pattern Discovery in Students' Evaluations of Professors: A Statistical Data Mining Approach

TL;DR

This paper applies advanced statistical data mining techniques to analyze student evaluations of professors, uncovering patterns such as the link between student dedication and evaluation scores.

Contribution

It introduces the use of multiple data mining methods to analyze evaluation data, revealing new insights into student feedback patterns.

Findings

Strong link between student attendance and evaluation scores

Identification of patterns relating student seriousness to evaluation outcomes

Use of diverse data mining techniques to analyze evaluation data

Abstract

The evaluation of instructors by their students has been practiced at most universities for many decades, and there has always been a great interest in a variety of aspects of the evaluations. Are students matured and knowledgeable enough to provide useful and dependable feedback for the improvement of their instructors' teaching skills/abilities? Does the level of difficulty of the course have a strong relationship with the rating the student give an instructor? In this paper, we attempt to answer questions such as these using some state of the art statistical data mining techniques such support vector machines, classification and regression trees, boosting, random forest, factor analysis, kMeans clustering. hierarchical clustering. We explore various aspects of the data from both the supervised and unsupervised learning perspective. The data set analyzed in this paper was collected from a university in Turkey. The application of our techniques to this data reveals some very interesting patterns in the evaluations, like the strong association between the student's seriousness and dedication (measured by attendance) and the kind of scores they tend to assign to their instructors.