Estudo comparativo do desempenho dos alunos de Ciência da Computação na UFC Quixadá usando mineração de dados

Abstract

This paper focuses on the comparative study of the performance of Computer Science students using data mining. In order to achieve better student performance, from 2015 onwards, the teachers of Programming Fundamentals held meetings to adopt a methodology consisting of assessment methods, a work and activity base. The evaluation methods consist of the application of continuous tests and partial evaluations by the online tool textit Moodle. In the Data Structure and Object Oriented Programming course, the solution was a collaboration between teachers and instructor students to apply exercise lists that help students practice and better understand the classes. The main objective of this paper is to compare student performance before and after using the assessment methodology in the Programming Fundamentals discipline. A comparison of results across different courses is made to assess the impact of such a methodology on student performance over the years 2013 to 2017. The data mining method called clustering was chosen to be used on university data. The textit K-means algorithm was applied to a data set of students from the Federal University of Ceará with the aid of the textit Scikit-Learn tool, which is an open source machine learning library in textit Python. With the results obtained from data mining, conclusions could be drawn about the effect of the adopted measures.