Visualising student satisfaction
Sam Cunningham, Mahsa Baktashmotlagh, Wageeh Boles · 2017
Student satisfaction is an important metric used in teaching and education, and is used in most educational institutions. There are many ways to measure student satisfaction (Elliott, 2002), however student satisfaction is often given as both a numerical score on a Likert scale as well as a text comment which contains further information. Although the numerical scores are often used, the free text comment is an invaluable source of information, often providing further pointers towards possible teaching enhancements. In this study, we explore the use of machine learning techniques to visualise student satisfaction. This visualisation will be exploited in the context of the following 2 research questions, 1. How can we use visual representations of comments to examine student satisfaction? 2. What impact can this have for educators? Using a dataset of over 20 subjects, many student comments and reviews were available for analysis. Primarily, using an analysis method called Latent Dirichlet Allocation (LDA) (Blei, et al., 2003), topics can be extracted from these comments about subjects. Sentiment analysis is then used to find the positivity and negatively of certain comments. Using the approach mentioned, comments from two subjects were analysed to demonstrate the capability of the process. For each subject, two figures were generated, one about the subject, and one focussing on a keyword or topic of interest. Both figures contain 9 automatically determined keywords as the focus of the plot. For each keyword, the length of the bar represents the frequency of keywords. The positive and negative sentiment is also represented on the same figure. From these figures, the reader can identify keywords related to a particular positive or negative sentiment. In this work, we have shown a process which automatically analyses and produces a visual representation for student satisfaction. The produced visuals can be used by lecturers, subject coordinators or managers to compare and review several subjects at once. As a lecturer, you would not need to know how this system works, but have access to student feedback. With the student feedback, these plots can be automatically generated. The visualisations can lead to quick detection of both positive and negative aspects within a subject, thus prompting appropriate action.