Code commit features before a help request show no correlation with interaction time, and commit-based scheduling strategies perform no better than FCFS¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1 · q2 · i0 negligible
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i0 Pearson correlations between commit frequency, last-commit elapsed time, last-commit LOC and interaction time are negligible and non-significant. → Z. Gao 2024 q2 i? Simulated commit-feature strategies yield no significant difference from FCFS on any queue metric. → Z. Gao 2024 (2)

Evidence¶

Z. Gao 2024¶

Z. Gao, G. S. de Oliveira, D. Babalola, C. Lynch, and S. Heckman. (2024). Who should i help next? simulation of office hours queue scheduling strategy in a cs2 course. Proceedings of the 17th International Conference on Educational Data Mining. https://doi.org/10.5281/zenodo.12729866

q2 · i0 · associational · r1

Correlational analysis of real request data (Table 2): commit frequency r = 0.0035 (p = 0.913), last commit elapsed time r = -0.0017 (p = 0.984), last commit LOC r = 0.0098 (p = 0.852) — "we could not find any correlation between the requests' interaction time and any code commit features".

"We tracked students' code commits before they raised a re-quest and while they waited in the queue and calculated the Pearson correlation of their code features and the interaction time for each request. Results are in Table 2, and we could not find any correlation between the requests' interaction time and any code commit features"

Z. Gao 2024 (2)¶

Z. Gao, G. S. de Oliveira, D. Babalola, C. Lynch, and S. Heckman. (2024). Who should i help next? simulation of office hours queue scheduling strategy in a cs2 course. Proceedings of the 17th International Conference on Educational Data Mining. https://doi.org/10.5281/zenodo.12729866

q2 · i? · causal · r1

Normal-load simulation repeated with three commit-feature strategies (highest commit frequency, lowest elapsed time since last commit, largest LOC). The results show no significant difference from FCFS on resolved requests, helped students, long wait requests, or overall wait time.

"the results show that compared with the FCFS strategy, using code commit features strategies does not yield any significant difference in resolved requests, helped students, long wait requests, and overall wait time"

Discussion¶

Related Claims¶

• The NSF advantage over FCFS appears only above an arrival-rate threshold (λ > 0.06); under relaxed queues no strategy makes a difference — related
• Under busy or normal queue load, the New Student First (NSF) strategy significantly increases the percentage of students who receive help compared with FCFS, LWF, and VLWF — related
• Choice of scheduling strategy does not significantly affect the number of resolved requests or students' overall wait time — related