I think that student performance prediction is one of the least interesting applications of learning analytics, but a lot of people are very interested in it. This paper (33 page PDF - 12 of text, the rest references) follows the very typical pattern of a systematic literature review (keyword search in commercial database, selection of relevant articles, filtering for quantitative studies). The study found more algorithms (89) being used than there were papers in the study (58). The authors also found "Another most used aspect to predict student performance is student demographics. Age, ethnicity, gender, housing, family history, and family socioeconomic level are among the student demographics." This is consistent with 25 years worth of studies I've reported here in OLDaily, and many more years before that. But if nobody's going to do anything about these disadvantages, what's the point of reporting on them?
Today: 1 Total: 120 [Direct link] [Share]
