This is a repository for research on how deprecation of Deep Learning libraries and APIs affects the quality of answers given on Stack Overflow.
The finds of this research is in overview.md.
- What is the deprecation status of the deep learning libraries in our study?
- Do users discuss the API versioning for the selected frameworks explicitly?
- What are the DL API versioning issues that are discussed on SO?
- How do Deep Learning API versioning issues affect the quality of answers in SO?
Do the users discuss the backward compatibility issues? [REF: An Empirical Analysis of Backward Compatibility in Machine Learning Systems]
Note that no relevant data has been included here at the instructions of the supervising professor.
All scripts can be run in a Python3 virtual environment. Be sure to install all
required packages via: pip install -r requirements.txt.
scripts/contains the scripts used to run the analysis. Scripts are organized by use case in directoriesdocumentation/contains the results of the analysis as of December 23, 2020.models/contains the lda, random forest, and linear regression modelsplots/contains any generated graphs
The datasets are not included in this repository.