This repository contains the replication package for "Are Fix-Inducing Changes a Moving Target? A Longitudinal Case Study of Just-In-Time Defect Prediction". The package contains the JIT data and scripts that are needed to reproduce the Figures 5–8 from the paper.

Required tools and packages

The following tools were installed on the machine where the scripts were originally executed:

• R version 3.2.3

In addition, the following R packages were installed:

• Formula 1.2-1
• Hmisc 3.17-2
• ggplot2 2.0.0
• lattice 0.20-33
• rms 4.4-2
• SparseM 1.7
• scales 0.3.0
• survival 2.38-3
• zoo 1.7-12

Configuring the Experimental Setting

The experiment has several configurable parameters. These can be changed by editing the scripts/experimental_setting.r file. The file contains the following options:

Option	Description
FILE_THRESH	Commits that touch more files than this value will be filtered out.
CHURN_THRESH	Commits that touch more lines than this value will be filtered out.
PROJECT_NAME	The name of the project that should be loaded from the data folder.
STRATA_PER_YEAR	A parameter to control the number of strata that should be created in the JIT data per year. Value of 2 (six months) and 4 (three months) are shown in the paper.
METRICS_FAMILIES	A list of families of metrics that are included in the fits of the JIT models.
FITTOOL	The function that is used to fit the models.
FITTOOL_PARMS	The parameters to be included with the FITTOOL function when it is called.
SPL	The function that should be used to fit splines in non-linear fits. Set to a NOOP function (included in the template file) if you want to (a) turn off nonlinear fits; or (b) try another FITTOOL that does not support splining.

Executing the Package

In order to run the package, execute:

Rscript scripts/replicate.r

Citing the Work

If you do make use of the dataset, please reference the original paper:

@article{mcintosh2018tse,
  Author = {Shane McIntosh and Yasutaka Kamei},
  Title = {{Are Fix-Inducing Changes a Moving Target? A Longitudinal Case Study of Just-In-Time Defect Prediction}},
  Year = {2018},
  Volume = {44},
  Number = {5},
  Journal = {IEEE Transactions on Software Engineering},
  Pages = {412--428}
}