Clustering Defect Reports
This is the companion website for the 2015 submission "A Text Clustering Approach to Predict Defect Resolution Time: A Replication and Beyond".
Below we provide the raw data needed to replicate the parts of our study conducted on open source software systems. Unfortunately, the defect reports and the corresponding metadata from Company A cannot be disclosed.
Further clarifications:
- For each defect report, the concatenated Title + Description is stored in a separate file. The file name contains the defect ID.
- All resolution times in the Eclipse metadata is in days.
- Clustering results from our experiments.
- The file Text_Clustering.rmp, containing all our experimental settings, can be imported into RapidMiner.
- The R script used to conduct our analysis and to plot our results.
For more information, please contact Markus Borg.
Android
Eclipse
Rapid Miner Process