← Back to all sparks
R

rainette

ANALYTICS
Velocity0.0

R package rainette by juba — release notes from GitHub.

rainette rebuilt its Reinert clustering in 0.2.0, tuned it in 0.3.0, and has coasted since.

text miningreinert methodclusteringshiny explorersr package
Current state
An R implementation of the Reinert textual clustering method, with interactive explorers for browsing clusters. The two substantive releases are behind it: 0.2.0 renamed the core segment-size arguments, fixed segment merging that had been crossing document boundaries, and added a document browser plus per-document cluster tables; 0.3.0 reworked the double classification in rainette2() with full and parallel arguments and much faster computation. The 2026 release is a vctrs compatibility fix plus a colors argument on rainette_plot().
Where it's heading
The package moved from correct-enough to trustworthy and then to maintained: results-changing fixes first, performance and options second, and now only upstream compatibility and small user-requested arguments. Wordcloud plots were flagged for deprecation in 0.3.0 and pulled from the explorers, narrowing the output surface rather than growing it. The same maintainer's questionr followed the same pattern in the same period.
Prediction
The deprecated wordcloud plot type is the obvious removal candidate, since it has carried a warning since 0.3.0 and has already been dropped from the interactive explorers.

Recent moves

  1. 7mo ago

    vctrs compatibility fix and custom cluster colors

    Restores compatibility with a new vctrs release and adds a colors argument to rainette_plot() for custom cluster colors. A user-contributed convenience alongside required upstream maintenance.

    View source ↗
  2. 3y ago

    R 3.6 palette compatibility and dendrogram fix

    Replaces a palette.colors() call with a manual vector for R 3.6 compatibility and fixes a dendrogram not displaying when the NA title is hidden. Two small fixes.

    View source ↗
  3. 4y ago

    Double classification reworked with restricted crossings and parallelism

    Reworks the double classification in rainette2() with a full argument to restrict crossings to maximal associations, an mclapply-based parallel option, and general speedups with progress bars. Wordcloud plots are marked for deprecation here.

    View source ↗
  4. 4y ago

    Merged segments visible in the document browser

    Adds an option to show merged segments in the document browser and fixes errors when a cluster's document-feature matrix is empty or contains empty-string features. Published a minute after 0.2.0 in the same backfilled batch.

    View source ↗
  5. 4y ago

    CRAN v0.2.0

    ⚡ SPARK

    The release that made results trustworthy: segments from different documents were being merged together, min_uc_size became min_segment_size with a new default of 0, and an interactive document browser plus per-document cluster tables arrived. Everything since builds on this shape.

    View source ↗