Results of IRJIT batched evaluation

These are results of preliminary study we conducted for our paper titled "IRJIT: A Simple, Online, Information Retrieval Approach for Just-In-Time Software Defect Prediction". IRJIT uses similarity between new and historical source code changes to identify bugs. By default, IRJIT collects and uses lines added to a source code change because SZZ algorithm only flags changes that introduce new lines as buggy [1]. Purushothaman et al. [2] also suggests that the probability of introducing a bug is higher with the addition of new lines. Hence, only lines added field might be enough to spot bugs. To confirm our assumption, we carried out an analysis where we use lines deleted for prediction. Our results which are provided here showed that lines deleted did not lead to a significant gain in the predictive performance of our model. Therefore, we only use the lines added field to build the IRJIT model.
[1] McIntosh, S., Kamei, Y.: Are fix-inducing changes a moving target? a longitudinal case study of just-in-time defect prediction. IEEE Transactions on Software Engineering 44(5), 412–428 (2017)
[2] Purushothaman, R., Perry, D.E.: Toward understanding the rhetoric of small source code changes. IEEE Transactions on Software Engineering 31(6), 511–526 (2005)