This package contains the datasets related to the manuscript

"An Empirical Study on the Usage and Availability of Machine Learning
Libraries in Open-Source Python Project"

authored by

Giulio Antoniol, Massimiliano Di Penta, Vittoria Nardone, and Cyrine Zid


The analyses have been performed during the year  2021.

In the following we describe the content of each file:

- all-projects-list.txt (all-projects.csv) -> list of all 4036 projects we analyzed
- projects-requirement-ML-passed.csv -> the set of 2082 projects successfully installing ML
      components

- full-install-for-the-2082-projects.csv the -> set of 2082  project requirements  installing log

- projects-with-requirements.csv  -> list of all 3421  projects with requirements.txt

- setup-only-projects.csv: list of the 580 setup.py files from the 490 projects with setup
  only setup
  
- all-requirement-deps-no-null.csv -> contains all rough  dependencies declared in requirement files
- all-setup-deps.csv -> contains all rough deps declared in projects that has ONLY a setup
    file and thus  no requirements.txt

- setup_py_install_log.csv -> detailed  log of setup.py installation

- chosen-requirements-paths.txt -> list of the 5917 requirement files
- manually-modified-requirements.csv -> manually modified requirements out of the 5917;
- notice 173 files failed the analysis due to non ASCII file names

- projects-with-setup.txt -> list of projects with setup.py
- current-deps.csv -> contains used libraries and corresponding version for any project (version declared into requirement files).
- dataset.csv -> contains all selected projects with their meta-data. The last column indicates the project type (i.e., whether it is a ML library or an APPLICATION)
- final_dephist.csv	-> contains the analysis of changes on libraries starting from the first commit.
- final_spreadness.csv -> shows the number of ML files/dirs and the total number of files per project.
- import-results.csv -> reports for every project file the imported libraries
- importDOT-results.csv -> reports for every project file the imported sub-module.

Furthermore in the directory ML-libraries-stats we report information about installed libraries and
versions are saved.