Textual Data Analysis in University Surveys to Understand and Reduce Student Dropouts

Abstract

In Argentine universities, the management of student data is a critical issue that needs to be addressed immediately.
These educational institutions collect a variety and quantity of data, such as the total number of students enrolled, the most chosen career/s, the dropout rate, among others. However, the retrieval, recording and analysis of these data is often inefficient and disorganized because many of them are in free textual con-tent format and come from diverse information sources. This abundance of data, while valuable, presents a significant challenge due to its unstructured and heterogeneous nature. That is, how to process textual Big Data to obtain information and then acquire knowledge that can help us make valuable decisions? In the educational domain, Text Analytics provides valuable information. This paper presents the Textual Data Analysis, collected from student surveys of two careers of the Faculty of Exact, Physical and Natural Sciences of the National University of San Juan. For this purpose, the ALCESTE method (Lexical Analysis of Cooccurrences in Simple Sentences of a Text) and other methods of the textual domain, such as word glossaries, concordances and the selection of the most specific vocabulary of each text, have been combined in order to provide a comparative tool. As a result, it is shown how the study of the distribution of the lexicon used in a text allows us to detect the structuring of the meanings present in it.