In the face of learning to program, students are often divided into two camps: those who excel and those who struggle. Through this challenge, some students manage to persevere from difficulty to ability. This study aims to identify the students who have made the most significant improvements in their programming skills, by leveraging various learning analytics and metrics. The field of learning analytics in programming has often sought to identify struggling students, utilizing metrics such as the error quotient (EQ) and time-on-task. This research distinguishes itself by taking a longitudinal approach to analyzing students’ programming data collected over a nine-month period. By framing error quotient through operant conditioning and time-on-task through expectancy-value theory, we aim to root these learning analytics in established learning theory to validate their relevance. These preliminary results illustrate how we can identify students who have improved the most based on these metrics.
