This repository holds the empirical analysis for the Linespots fault-prediction algorithm. The Linespots reference implementation and evaluation code can be found here.
The analysis reports for the recent An empirical study of Linespots: A novel past-fault algorithm paper can be found here:
The csv file with the evaluation output, in case you want to run your own analysis from the ground up can be found here or can be replicated using the docker image.
My master thesis about Linespots: https://hdl.handle.net/20.500.12380/300418. There is no analysis report and no reproducability package per se but the evaluation data can be found here.
While not perfect, this should mostly reproduce the results:
- Clone https://gitlab.com/sims1253/linespots-lib and checkout the
sims/evaluation
branch. - Run
pipenv install
in the repository - Run
pipenv run project_getter.py
from thelinespots/utils
directory - Run
pipenv run json_builder.py
from thelinespots/utils
directory - Change the name of the json config in
full-evaluation.py
(line 44) tofinal-config.json
- Run
pipenv run full-evaluation.py
from the top-level directory of the repository - This should create a
full_evaluation.csv
file in theevaluation_projects
directory
I am aware that this process is error prone (and won't reproduce the exact results due to how I calculated commit ranges) but it I do not intend to fix this. For an up to date analysis see the paper above.
The beginning of the Linespots Journey. Published here(pdf). I never did a proper statistical analysis for this but the code is available here and the repository contains some instructions on how to reproduce what I did.