No tests required

Abstract

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for personal research or study, educational, or not-for-prot purposes provided that:• a full bibliographic reference is made to the original source • a link is made to the metadata record in DRO • the full-text is not changed in any way The full-text must not be sold in any format or medium without the formal permission of the copyright holders.Please consult the full DRO policy for further details. ABSTRACTResearch over the past fifty years into predictors of programming performance has yielded little improvement in the identification of at-risk students. This is possibly because research to date is based upon using static tests, which fail to reflect changes in a student's learning progress over time. In this paper, the effectiveness of 38 traditional predictors of programming performance are compared to 12 new data-driven predictors, that are based upon analyzing directly logged data, describing the programming behavior of students. Whilst few strong correlations were found between the traditional predictors and performance, an abundance of strong significant correlations based upon programming behavior were found. A model based upon two of these metrics (Watwin score and percentage of lab time spent resolving errors) could explain 56.3% of the variance in coursework results. The implication of this study is that a student's programming behavior is one of the strongest indicators of their performance, and future work should continue to explore such predictors in different teaching contexts.

Transfer-Learning Methods in Programming Course Outcome Prediction

“…Both traditional machine-learning methods and transfer-learning methods show reasonable performance in determining whether the student will pass the course. This result is in line with previous studies that have identified the possibility of assessing students' performance based on various process-and background-based metrics [1,19,22,32,34,38]. However, when comparing our study to the previously mentioned ones, the difference in context was significantly larger, and our study had rather generalized features.…”

“…Error quotient and its relatives (see, e.g., References [2,5,38]) are approaches where the data used for the outcome prediction has required significant insight into the programming process and the internals of the used programming language. This is not necessary, however, as one can also use higher level information such as the number of actions or attempts that a student makes on assignments as a feature [1,6].…”

“…The working process data were augmented with the background survey details. The background survey was constructed in house and inspired by the surveys in References [1,38]. Some details such as whether the student has Computer Science as a major were included due to our specific context-in the studied context, students pick their major before entering the University, and as such, choosing Computer Science as a major could indicate motivation.…”

Predicting Student Failure in an Introductory Programming Course with Multiple Back-Propagation

“…Most of these are based on analysis of student interaction with the programming language compiler, such as: investigating novice programming mistakes [1]; exploring compilation behavior [16]; analyzing response to compiler error messages [23]. Other studies focus on comparing and testing traditional predictors of performance and new datadriven predictors [27]. Several works have found that the mathematical ability and exposure to mathematics courses are important predictors of performance on introductory programming courses [25,28,30].…”

50 Years of CS1 at SIGCSE

“…In 2017, Kirkpatrick et al introduced an alternative CS1 for students with prior programming knowledge [25]. Predicting & measuring success [6,46] and research on Programming process data [2] have also only appeared since the turn of the century. It is interesting to note that concerns about large enrollments appeared as early as 1974 [11], while Retention, dating back as far as 1984, has shown significant interest more recently [34].…”