Predicting students' grades in computer science courses based on complexity measures of teacher's lecture notes

In the past, predicting students grades has been done primarily by looking at students past test scores, at their history [1, 2], by looking at their current notes and/or their note taking ability [3, 5], from surveys about various other aspects of their background, or by a combination of such factors [4, 6, 7]. In this paper a different approach is taken -- here, the complexity of the teacher's lecture notes is examined as a predictor of the students' grades. The results of this research indicate that simple measures such as total number of words in the lecture or total number of lines of code that appear in the lecture are not good predictors of students' grades; however, "buzzword density," or the total number of words with Computer Science meaning divided by the total number of words in the lecture does predict students' grades.