Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
55 lines (48 loc) · 4.05 KB

CONTRIBUTING.md

File metadata and controls

55 lines (48 loc) · 4.05 KB

How To Contribute

There are perhaps two different paths, based on your personal preferences:

  • A. Coding: Get into the code and get to a small pull request ASAP, or
  • B. Learning: Spend some time learning the context and ecosystem of the project.

Both are great approaches, depending on whether you like working inside-to-out (actions, then context) or outside-to-in (context, then action).

A. Action: Code First

Small things we need done: (these will move to pull requests soon)

  • some simple tests of existing functionality
  • CI running on GitHub actions, running a basic install on a few versions of Python
  • a simple documentation website using mkdocs (e.g., front-page from README + API page)
  • write a dataset processor for importing representative likert survey data into valency-anndata
  • add PaCMAP algorithm support (model off of scanpy.tl.umap())
  • add LocalMAP algorithm support (model off of scanpy.tl.umap())
    • despite the claims of PaCMAP's README, LocalMAP is already merged into PaCMAP's codebase
  • val.viz.schematic_diagram: remove misleading grid lines from all non-X blocks
  • val.datasets.polis.translate_statements: make is_translated more clever (to know when things AREN'T translated)
  • val.datasets.polis: summarize changes during .load()
  • val.viz.schematic_diagram: when no diff_from arg provided, assume diff from empty AnnData (#4)
  • val.datasets.polis: when above n participants, require allow_large_scrape=True to use convo url
  • val.datasets.polis: rename adata.uns objects to have raw_ prefix
  • val.tools.recipe_polis: document polis pipeline, e.g. flowchart
  • val.tools.impute: convert to use scikit-learn's SimpleImputer
  • val.tools.polis: Add helper method to extract metadata statements into obs
    • clustering doesn't happen on these, as these votes get zeroed out in zero_mask
  • val.tools.kmeans: get kmeans++ init strategy working (in red-dwarf)
  • val.viz: re-export sc.pl.embedding into valency-anndata
  • research new tools to assess and validate cluster labels
  • print Creative Commons attribution when loading Polis data, as per report page note.
  • val.preprocessing.calculate_qc_metrics: warn when performed on non-sparse data (expects NaNs for seen/unseen vote metrics to make sense)

See TODO.md or ask patcon for other ideas!

B. Learning: Context First