LoginSign Up
← Back to all sparks
T

tidyr

ANALYTICS
Velocity0.0

Tools for creating tidy data in R

github.com/tidyverse/tidyr ↗

tidyr replaced separate() with a family that says what it does.

tidyversedata-reshapingpivotingapi-designr-language
◆Current state
tidyr is at 1.3.2, a collection of argument additions — fill() gains .by, expand_grid() gains .vary — and better error messages around unchop() and pivot_wider_spec(). The structural work is 1.3.0, which introduced separate_wider_delim(), separate_wider_position(), separate_wider_regex(), separate_longer_delim() and separate_longer_position() as thorough replacements for separate(), extract() and separate_rows().
◆Where it's heading
Two habits define this window. Verbs are being split into explicitly named variants rather than overloaded with arguments, which is what the separate_* family does to separate(). And .by is spreading as the standard way to express grouping inline — nest(.by=) in 1.3.0, fill(.by=) in 1.3.2 — pulling users away from wrapping calls in group_by().
◆Prediction
Given that .by has now reached fill() and nest(), the next release most likely extends the same argument to further verbs rather than reworking another function family.

◆Recent moves

  1. 7mo ago

    fill() gains .by; expand_grid() gains .vary

    fill() gains .by for inline grouping and expand_grid() gains .vary to control which column varies fastest. Several error messages improve and unite() now returns an empty-string column instead of erroring on an empty selection.

    View source ↗
  2. 2y ago

    pivot_wider duplicate message uses modern syntax

    A single change modernising the syntax in pivot_wider's duplicate-detection helper message. Cosmetic.

    View source ↗
  3. 3y ago

    separate_wider_* family supersedes separate() and extract()

    Introduces the separate_wider_* and separate_longer_* family, superseding separate(), extract() and separate_rows() with faster, more consistent functions and a defined approach to malformed input. nest() gains .by, and pivot_longer() gains cols_vary.

    View source ↗
  4. 3y ago

    Hot patch for R CMD check failures

    A hot patch for R CMD check failures, with no other content.

    View source ↗
  5. 4y ago

    pivot_wider() gains names_expand, id_expand and unused_fn

    pivot_wider() gains names_expand and id_expand for making implicit missing combinations explicit, names_vary for column ordering, and unused_fn for summarising columns left out of the pivot. Note the stored body is cut at 8000 characters, so later items in this changelog are not visible here.

    View source ↗
  6. 4y ago

    Large speedups to unchop(), unnest() and expand_grid()

    Performance work with wide reach: unchop() gets much faster and that propagates through unnest(), unnest_longer(), unnest_wider() and separate_rows(), while expand_grid() roughly doubles in speed. Mixing vectors and data frames in an unnested list-column now errors.

    View source ↗