LoginSign Up
← Back to all sparks
D

datawizard

ANALYTICS
Velocity0.0

Data wrangling, cleaning and statistical transformation toolbox for R

github.com/easystats/datawizard ↗

datawizard is turning easystats' data layer into a general-purpose I/O and reshaping tool

data-wranglingeasystatsfile-formatsbreaking-changesr-stats
◆Current state
datawizard handles the data preparation half of the easystats stack — reshaping, recoding, describing, and reading and writing files. The 1.x releases have pushed hardest on I/O: parquet via nanoparquet, then password-protected R formats, alongside a run of breaking cleanups in data_to_wide(), data_modify() and describe_distribution().
◆Where it's heading
The package is willing to break its own interfaces to reach behavior users expect from tidyr and friends — data_to_wide() explicitly moved toward pivot_wider() semantics, and data_modify() stopped guessing whether a string was an expression. Output formatting is consolidating behind insight's display() and tinytable. The direction is fewer surprises and more file formats, not more statistics.
◆Prediction
Expect encryption and format support to extend past R-native files if it continues, and further alignment of print and display behavior with the shared insight infrastructure.

◆Recent moves

  1. 3mo ago

    Encrypted data files via a password argument on read/write

    AES-GCM encryption arrives in data_read() and data_write(), currently limited to R-native formats — a new capability for the I/O side that has been this package's most active surface. data_summary() also gains multi-value expressions.

    View source ↗
  2. 10mo ago

    data_to_wide() moves toward pivot_wider() semantics

    Retires the broken values_fill argument and stops dropping empty columns after widening, closing the behavior gap with tidyr. Also adds as.prop.table() and tinytable output, continuing the display consolidation.

    View source ↗
  3. 1y ago

    Parquet read and write support via nanoparquet

    The first of the significant I/O additions in this window, plus display() and as.table() methods for data_tabulate() that make frequency tables usable in downstream tests.

    View source ↗
  4. 1y ago

    data_modify() stops inferring expressions from strings

    The heaviest breaking release here: string arguments are treated as values unless wrapped in as_expr(), describe_distribution() changes its grouped and interval output, and data_read() stops coercing Bayesian models into data frames. Predictability over convenience.

    View source ↗
  5. 1y ago

    datawizard 1.0.2

    A numerical-tolerance fix for CRAN check platforms. No user-facing change.

    View source ↗
  6. 1y ago

    data_arrange() preserves single-column data frames

    Two small correctness fixes, one of them for an R-devel change in grep(). Routine upkeep.

    View source ↗