← Back to all sparks
F

feasts

ANALYTICS
Velocity0.0

Feature extraction and statistics for time series analysis in R

feasts is splitting itself in two, moving every plot into ggtime

time-seriesr-statsdeprecationpackage-splitvisualizationtidyverts
Current state
feasts provides feature extraction and statistics for tsibble time series. The last two releases are dominated by one decision: all of its gg_*() plotting functions are being moved out into a separate ggtime package. 0.4.2 announced the deprecation and 0.5.0 makes ggtime a dependency with soft-deprecation messages on every re-export.
Where it's heading
The package is narrowing to its stated purpose — features and statistics — and shedding graphics entirely over a deliberately slow two-year window. Everything else in the recent history is ggplot2 compatibility work and narrow seasonal-plot bug fixes, which is consistent with a maintainer trimming surface area rather than growing it.
Prediction
The next releases should be compatibility upkeep while the ggtime deprecation runs its course; the re-exports stay until the announced window closes.

Recent moves

  1. 6mo ago

    feasts moves its plots to ggtime behind a 2-year deprecation

    ⚡ SPARK

    Executes the split announced in 0.4.2: the graphics functions now live in ggtime and feasts only re-exports them with a deprecation message. This is the release where the boundary becomes real for users.

    View source ↗
  2. 11mo ago

    ggplot2 4.0.0 compatibility and the ggtime deprecation notice

    Prepares for ggplot2 4.0.0 and adds a plot_type argument to gg_tsresiduals(), but its lasting content is the announcement that all ggplot2 functionality will move to ggtime. The move itself lands in 0.5.0.

    View source ↗
  3. 1y ago

    gg_season() fix for sub-weekly daily data

    A single seasonal-plot fix for daily data with seasonality longer than a week. Narrow scope, no interface change.

    View source ↗
  4. 1y ago

    Impulse-response plots and Johansen cointegration tests

    Adds gg_irf() to visualise the impulse responses fable started producing, plus two cointegration tests borrowed from urca. A direct complement to fable's multivariate model work in the same period.

    View source ↗
  5. 2y ago

    Patch for ggplot2 3.5.0 breaking changes

    Compatibility patch for ggplot2 3.5.0, with a small correction to how seasonally adjusted series are computed from classical decomposition.

    View source ↗
  6. 3y ago

    CRAN patch for S3 method consistency

    A CRAN-check patch with no user-visible effect.

    View source ↗