6.8 KiB
6.8 KiB
progress:: {{renderer :todomaster}}
icon:: 📅
boox-notes::
- ### Tasks
-
- ### Notes
-
- #Clip 07-07-2025 09:32 Posta - Davide Di Ruscio - Outlook
merging RQ3.C into RQ3.A.
- Add a threats to validity, do the hyperparameter tuning for MLP and check if things change.
- #Clip 07-07-2025 09:32 Posta - Davide Di Ruscio - Outlook
The conclusions of this paper are highly dependent on the quality of the test cases.
- Threats to external validity
- #Clip 07-07-2025 09:32 Posta - Davide Di Ruscio - Outlook discussing (or, if possible, evaluating) additional performance-testing frameworks (e.g., Gatling, JMeter) or load/stress tests to broaden the applicability of your findings.
- #Clip 07-07-2025 09:32 Posta - Davide Di Ruscio - Outlook elaborate on how JMH's methodological constraints (e.g., microbenchmark isolation) might bias coverage and cost measurements, and whether your results would generalize to other performance testing styles.
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook acknowledge how method-level results might differ from statement-level insights.
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook consider prototyping a minimal plugin or outlining its design,
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook discuss integrating dynamic profiling metrics
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook reflect on how dynamic features might improve predictive accuracy beyond the top F1≈0.47
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook justify these choices, or report any preliminary experiments that guided them
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook sketch a roadmap for building a "performance-coverage" plugin—what metadata it would collect, how it would integrate with build tools, and how teams might act on its reports.
- #Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook
It may make more sense to focus on deep learning models for prediction.
- Future work
-
#Clip 07-07-2025 09:33 Posta - Davide Di Ruscio - Outlook If the prediction is accurate, the recommended method should be performance-critical, yes?