Identifying the Danger Zones: Predictors of Success and Failure in a CS1 Course
Nathan Rountree, Janet C. Rountree, Anthony V. Robins · 2001
We present the results of a survey which focuses on the backgrounds and expectations of a group of CS1 students in the first weeks of semester. When comparing their survey answers to their final grades on the course, we saw some surprising things: the group which indicated an intention to continue in computer science did no better than any other, and the strongest single indicator of success seems to be "expecting to get an A from the course." In order to see if there were any particular combinations of answers that indicated success or failure, we ran a decision-tree classifier over the survey data. This resulted in the identification of differing "danger zones" for each year of study, in which we observed about twice the expected proportion of failing students. 1.