Milestone 2 Meeting Minutes

[cmmclaug]

March 1st 2021 Milestone 2 Kickoff:

• Make sure Python package is usable by end of milestone
• Revise functions first
• Write tests before functions
• Iterate functions/tests
• Neel/Daniel to set up R repo
• Wait until lecture before assigning all issues
• Aim to have Python test and function beta for Thursday
• Create the R repo structure first before working on R tasks
• We should continue to have early group kickoff meetings
• Update issues with notes when creating PRs

[cmmclaug]

March 4th 2021 Lab Discussion:

• During function revision and discussion with Tiffany, we came to the following clarification on function purposes:
  • Our regression functions will train sklearn models and report important statsmodel model values
  • Not enough time for PostLasso inference
  • We do not have to create a completely unique package
• New high-level directive: Create wrapper functions that produce a fit sklearn model and an output of statistical model and feature summaries from statsmodel model.
• Model summary metrics to report:
  • N observations
  • Loss function used
  • Adjusted R-squared
  • F-test probability
• Report feature coefficient summary table as a dataframe
• Add a missing data plot to the EDA function

[cmmclaug]

March 5th 2021 Group Check-in Meeting:

• Use error_string constants file to keep DRY
• Change data_frame inputs to df
• arid_linreg() is now outputting an sklearn and statsmodel models and outputting some summary info
• Remember that commit messages on R repo are subject to grading
• Make sure to use GitHub Flow workflow on R repo; always branch and create PRs from-to main
• R package will create an sklearn interface in R environment.