This repo contains the appendix/instruments for our paper, "Machine Learning Experiment Management Tools: A Mixed-Methods Empirical Study".
Survey - The instruments/appendix for our survey include:
- Questionnaire: Content of the Google form used as questionnaire for the survey.
- Responses: A collection of data elicited from 3 different batches of participants. Some information such as links and participants' emails have been redacted to preserve anonymity.
- QA codebook: A breakdown of the code extraction for qualitative analysis on the open-ended questions.
Controlled experiment - The instruments/appendix used for our controlled experiment include:
- Neptune.ai-material: This directory includes a short Neptune tutorial, the initial python scripts for the experiment phases using Neptune.ai to track ML assets, and instruction files for each of the participant group.
- DVC-material: This directory includes a short DVC tutorial, the initial python scripts and DVC config files for the experiment phases that used DVC to track and manage ML assets, and instruction files for each of the participant group
- No-Tool-material: This directory includes the initial python scripts for the experiment phases, which participants carried out without any supporting tool, and the instruction files for each of the participant group
- Questionnaire & Responses:
- Questionnaire: Content of the Google Form used as a questionnaire for the controlled experiment. Some information such as links and participants' emails have been redacted to preserve anonymity.
- Response: A collection of data obtained from the controlled experiment as responses to our questionnaire
- Statistical Data Analysis: Raw data, extracted data and statistical test analysis of quantitative data points.