CRAN v0.4.1

Minor patch for upcoming release of ggdist v3.3.1

Improvements

• Added (scaled) pinball loss metrics to interval_accuracy_measures (#379).
• Improved use of random seed in parallel modelling and forecasting (#384).
• Documentation improvements

CRAN v0.4.0

Improvements

• Improved handling of combination_model() when used with transformed
  component models.
• autoplot(<fbl_ts>), autolayer(<fbl_ts>) and autoplot(<dcmp_ts>) now use
  the ggdist package visualising uncertainty with distributional vectors.

CRAN v0.3.4

fabletools 0.3.4

New features

• The formula parser now identifies and stores length 1 values in the
  transformation environment. This simplifies common tasks like automatic
  box-cox parameters for each series, which can now be done with
  fable::ARIMA(box_cox(y, feasts::guerrero(y))).

Improvements

• Added support for visualising different point forecasts (say means and medians)
  when only one forecast is to be plotted for each series.

Bug fixes

• Resolved issue with autoplot(<fbl_ts>) not identifying multiple point
  forecasts by linetype.
• Fix for indexing of bottom series in top_down() and middle_out()
  reconciliation methods (#362, #364 @FedericoGarza)

CRAN v0.3.3

fabletools 0.3.3

Improvements

• Fixed handling of transformed distributions which accept a parameter from the
  dataset.
• . in a model formula for xreg implemented with special_xreg() will now
  include all measured variables (excluding the index and key variables).
• Improved handling of transformations with forecast sample distributions.
• Added support for reconciling sample paths.
• accuracy(<fbl_ts>) can now summarise accuracy over key variables. This is
  done by specifying the accuracy by argument and not including some (or all)
  of the fable's key variables (#341).
• Like forecast(), generate() will now keep exogenous regressors in the
  output table.
• Re-export generics::forecast() for better compatibility with registering
  methods alongside other packages (#375).

CRAN v0.3.2

fabletools 0.3.2

New features

• Added hypothesize() generic for running statistical tests on a trained model.
• Added combination_weighted() function for producing a combination model with
  arbitrary weights.

Improvements

• The fallback residuals() method now handles transformations when
  type = "innovation".
• Improved supported expressions for producing combination models. The
  appropriate response variable is now simplified for all functions that produce
  that original response variable. This notably includes 0.7*mdl1 + 0.3*mdl2 -
  if mdl1 and mdl2 are models with the same response variables, then the
  resulting combination model will also have the same response variable.
• Documentation improvements.

Bug fixes

• Fixed issue with exogenous regressors (xreg) in reconciliation methods that
  partially forecast the hierarchy.
• Fixed issue with keys being dropped when several mdl_df (mable) objects were
  combined.

CRAN v0.3.1

New features

• Added outliers() generic for identifying the outliers of a fitted model.
• Added special_xreg() special generator, for producing a model matrix of
  exogenous regressors. It supports an argument for controlling the default
  inclusion of an intercept.
• Migrated common_xregs helper from fable to fabletools for providing a
  common and consistent interface for common time series exogenous regressors.
• Added experimental support for passing the tsibble index to features()
  functions if the .index argument is used in the function.

Improvements

• Added transformation support for fallback fitted(h > 1) method (#302).
• Documentation improvements.

CRAN v0.3.0

New features

• Added scenarios() function for providing multiple scenarios to the
  new_data argument. This allows different sets of future exogenous regressors
  to be provided to functions like forecast(), generate(), and
  interpolate() (#110).
• Added quantile_score(), which is similar to percentile_score() except it
  allows a set of quantile probs to be provided (#280).
• Added distribution support for autoplot(<dable>). If the decomposition
  provides distributions for its components, then the uncertainty of the
  components will be plotted with interval ribbons.
• Added block bootstrap option for bootstrapping innovations in generate().
• Added multiple step ahead fitted values support via fitted(<mable>, h > 1).
• Added as_fable(<forecast>) for converting older forecast class objects to
  fable data structures.
• Added top_down(method = "forecast_proportion") for reconciliation using the
  forecast proportions techniques.
• Added middle_out() forecast reconciliation method.
• Added directional accuracy measures, including MDA(), MDV() and MDPV()
  (#273, @davidtedfordholt).
• Added fill_gaps(<fable>).

Improvements

• The pinball_loss() and percentile_score() accuracy measures are now scaled
  up by 2x for improved meaning. The loss at 50% equals absolute error and the
  average loss equals CRPS (#280).
• Automatic transformation functions formals are now named after the response
  variable and not converted to .x, preventing conflicts with values named .x.
• box_cox() and inv_box_cox() are now vectorised over the transformation
  parameter lambda.
• RMSSE() accuracy measure is now included in default accuracy() measures.
• Specifying a different response variable in as_fable() will no longer
  error, it now sets the provided response value as the distribution's new
  response.
• Minor vctrs support improvements.

Bug fixes

• Data lines in fable autoplot() are now always grouped by the data's key.
• Fixed bottom_up() aggregation mismatch for redundant leaf nodes (#266).
• Fixed min_trace() reconciliation for degenerate hierarchies (#267).
• Fixed select(<mable>) not keeping required key variables (#297).
• Fixed ... not being passed through in report().

CRAN v0.2.1

New features

• Added bottom_up() forecast reconciliation method.
• Added the skill_score() accuracy measure modifier.
• Added agg_vec() for manually producing aggregation vectors.

Improvements

• Fixed some inconsistencies in key ordering of model accessors (such as
  augment(), tidy() and glance()) with model methods (such as forecast()
  and generate()).
• Improved equality comparison of agg_vec classes, aggregated values will now
  always match regardless of the value used.
• Using summarise() with a fable will now retain the fable class if the
  distribution still exists under the same variable name.
• Added as_fable.forecast() to convert forecast objects from the forecast
  package to work with fable.
• Improved CRPS() performance when using sampling distributions (#240).
• Reconciliation now works with hierarchies containing aggregate leaf nodes,
  allowing unbalanced hierarchies to be reconciled.
• Produce unique names for unnamed features used with features() (#258).
• Documentation improvements
• Performance improvements, including using future.apply() to parallelize
  forecast() when the future package is attached (#268).

Breaking changes

• The residuals obtained from the augment() function are no longer controlled
  by the type argument. Response residuals (y - yhat) are now always found
  in the .resid column, and innovation residuals (the model's error) are now
  found in the .innov column. Response residuals will differ from innovation
  residuals when transformations are used, and if the model has non-additive
  residuals.
• dist_*() functions are now removed, and are completely replaced by the
  distributional package. These are removed to prevent masking issues when
  loading packages.
• fortify(<fable>) will now return a tibble with the same structure as the
  fable, which is more useful for plotting forecast distributions with the
  ggdist package. It can no longer be used to extract intervals from the
  forecasts, this can be done using hilo(), and numerical values from a
  <hilo> can be extracted with unpack_hilo() or interval$lower.

Bug fixes

• Fixed issue with aggregated date vectors (#230).
• Fixed display of models in View() panel.
• Fixed issue with combination models not inheriting vctrs functionality (#237).
• aggregate_key() can now be used with non-syntactic variable names.
• Added tsibble cast methods for fable and dable objects, fixing issues with
  tidyverse functionality between datasets of different column orders (#247).
• Fixed refit() dropping reconciliation attributes (#251).

CRAN v0.2.0

New features

• Distributions are now provided by the distributional package, which is more
  space efficient and allows calculation of distributional statistics including
  the mean(), median(), variance(), quantile(), cdf() and density().
• autoplot.fbl_ts() and autolayer.fbl_ts() now accept the point_forecast
  argument, which is a named list of functions that describe the method used to
  obtain the point forecasts. If multiple are specified, each method will be
  identified using the linetype.
• Added accuracy measures: RMSSE(), pinball_loss(), scaled_pinball_loss().
• Added accessor functions for column names (or metadata) of interest. This
  includes models in a mable (mable_vars()), response variables
  (response_vars()) and distribution variables (distribution_var()).
• Added support for combinations of non-normal forecasts, which produces mean
  point forecasts only.
• Added support for reconciling non-normal forecasts, which produces reconciled
  point forecasts only.

Improvements

• Improved dplyr support. You can now use bind_*() and *_join() operations
  on mables, dables, and fables. More verbs are supported by these extension
  data classes, and so behaviour should work closer to what is expected.
• Progress reporting is now handled by the progressr package. This allows you to
  decide if, when, and how progress is reported. To show progress, wrap your
  code in the progressr::with_progress() function. Progress will no longer be
  displayed automatically during lengthy calculations.
• Improved support for streaming data to models with transformed response
  variables.
• hilo.fbl_ts() now keeps existing columns of a fable.
• forecast() will now return an empty fable instead of erroring when no
  forecasts are requested.
• is_aggregated() now works for non-aggregated data types.
• Documentation improvements.

Breaking changes

• The fable returned by forecast() now stores the distribution in the column
  named the response variable (previously, this was the point forecast). Point
  forecasts are now stored in the .mean column, which can be customised using
  the point_forecast argument.
• The bias_adjust option for forecast() is replaced by point_forecast,
  allowing you to specify which point forecast measures to display (fable/#226).
  This has been done to reduce confusion around the argument's usage,
  disambiguate the returned point forecast's meaning, and also allow users
  to specify which (if any) point forecasts to provide.
• The data coercion functions as_mable, as_dable, and as_fable have been
  changed to accept character vectors for specifying common attributes (such as
  response variables, and distributions).
• The models argument for mable and as_mable has been replaced with model
  for consistency with the lack of plural in key.
• Intervals from multivariate distributions are now returned as data frames of
  hilo intervals. The columns are the response variables. Similar structures
  are returned when computing other distributional statistics like the mean.
• hilo intervals can no longer be unnested as they are now stored more
  efficiently as a vctrs record type. The unpack_hilo() function will continue
  to function as expected, and you can now obtain the components of the interval
  with x$lower, x$upper, and x$level,
• rbind() methods are deprecated in favour of bind_rows()
• The row order of wide to long mable operations (such as accuracy()) has
  changed (due to shift to pivot_longer() from gather()). Model column name
  values are now nested within key values, rather than key values nested in
  model name values.

Bug fixes

• Fixed show_gap option not working when more than one forecast is plotted.
• Fixed autolayer() plotting issues due to inherited aesthetics.
• aggregate_key() no longer drops keys, instead they are kept as .
• Forecast reconciliation now works with historical data that is not temporally
  aligned.
• Fixed forecast() producing forecasts via h when new_data does not
  include a given series (#202).

CRAN v0.1.3

Improvements

• Better support for tidyverse packages using vctrs.
• Performance improvements for reconciliation and parsing.
• xreg() can now be called directly as a special.

Bug fixes

• Fixed accuracy.fbl_ts() error when certain names were used in the fable.