My SciELO
Services on Demand
Journal
Article
Indicators
- Cited by SciELO
Related links
- Similars in SciELO
Share
Conrado
Print version ISSN 2519-7320On-line version ISSN 1990-8644
Abstract
GUANIN-FAJARDO, Jorge; CASILLAS, Jorge and CHIRIBOGA-CASANOVA, Washington. Semi-supervised learning to discover the average scale of graduation of university students. Conrado [online]. 2019, vol.15, n.70, pp.291-299. Epub Dec 02, 2019. ISSN 2519-7320.
Institutions of higher education omit to a certain extent the factors that delay the rates of promotion of university students. The delay cannot always be disclosed due to the diversity of study programs, from the beginning of the career to the completion of the program and graduation. This paper used the student data set for 5 full academic years (grades 1-5), 53 variables, and 849 observations of different university careers. Thus, variables were explored and data mining with semi-supervised learning techniques was used to discover associations that detect graduation categories of students. Therefore, the rules of interest were discovered using the metrics of support, confidence and elevation of the rules of association. The findings suggest that the ages of the group of teachers between second and third year, as well as the grade point average between courses and the employability of students, are the main factors influencing the graduation rates of university students.
Keywords : Semi-supervised learning; educational analysis; data mining; higher education; graduation time.