Modelagem e Predição de Reprovação de Acadêmicos de Cursos de Educação a Distância a partir da Contagem de Interações
Year: 2015
Type: article
Abstract: The high students' failure and dropout rates are often pointed as two of the main obstacles for the implementation and consolidation of Distance Learning courses. The early identification of at risk students can decisively aid teachers and assistants. The present work describes results of the application of machine learning techniques to perform this task using only the count of interactions as attributes. The experiments demonstrated the feasibility of using only the counts of students interaction to generate reasonably accurate predictions, and that introducing attributes derived from the counts (e.g. means) is useful for more accurate predictions when data is sparse. Moreover, it was also demonstrated that Bayesian networks are adequate for the problem and that it is possible to train models in different sources of examples, such as between different groups and semesters. The present approach is virtually applicable in any scenario where it is possible to count interactions of any kind.
Institution: Universidade Federal de Pelotas
Language: Portuguese
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W2557716502
