Nothing
residuals.hybridModel()
and added tests.Makefile
targets for building and package development.rolling
argument to hybridModel()
that can be used when weights = "cv.errors"
to control the rolling
argument in cvts()
.comb
as an argument to thiefModel()
.accuracy()
formals to support changes in the "forecast" package version 8.12.thief()
function can now be created with the new thiefModel()
function. The API is similar to that of hybridModel()
.cvts()
. This results in significantly faster execution and less memory usage, particularly when the FUN
and FCFUN
functions are very quick (e.g. snaive()
, rwf()
, stlm()
), the time series is short, few cores are used, or few CV folds run.cvts()
examples.cvts()
.xreg
argument passed in should now be a matrix instead of a dataframe for consistency with "forecast" v8.5.hybridModel
objects that use far less memory and that print more cleanly to the console. For example, previously hm <- hybridModel(wineind); format(object.size(hm), units = "auto")
produced a 5.8 Mb object but now it is only 314.8 Kb.hybridModel()
. This can be controlled by setting parallel = TRUE
and setting num.cores
. By default this is not enabled since the performance improvement typically only occurs when fitting auto.arima
and tbats
models on long series with large frequency (e.g. taylor
).z.args
for the snaive()
model.tbats()
and snaive()
models now respect and use lambda
when passed in t.args
and z.args
.snaive
model are now handled correctly.inst/davidshaub@gmx.com.key
and hosted on both GitHub and GitLab in pkg/inst/davidshaub@gmx.com.key
.PI.combination
argument to forecast.hybridModel()
. The default behavior is to follow the existing methodology of using the most extreme prediction intervals from the component models. When "mean"
is passed instead, a simple (unweighted) average of the component prediction intervals is used instead.snaive()
model to the ensemble. It is disabled by default, but can be added with "z".cvts()
for the FCFUN
argument: custom forecasting functions should now return a S3 "forecast" object with the point forecast in $mean
, and the ts
properties should be properly set.cvts()
now defaults to 2 corescvts()
to the vignette.cvts()
introduced in version 1.0.8 when a custom FUN
or FCFUN
is used that requires packages other than "forecast" or "forecastHybrid".thetam()
function now checks for an input time series with less length than the seasonality. Similarly, hybridModel()
detects this behavior. Thanks to Nicholas Fong for the bugfix.cvts()
usage example in documentation for "GMDH".forecast.hybridModel()
when for models where xreg
was not supplied to all of arima/nnetar models.ts
objects created with the "timekt" package can now be used in hybridModel()
.doParallel
and forecast
packages are now imported instead of loading their entire namespaces.cvts()
now supports parallel fitting through the num.cores
argument.
Note that if the model that you are fitting also utilizes parallelization,
the number of cores used by each model multiplied by num.cores
passed to
cvts()
should not exceed the number of cores on your machine.MAJOR.MINOR.RELEASE_NUM
.ggplot2
namespace, only specific functions are now imported.accuracy()
, so this is imported and no longer declared in "forecastHybrid".cvts()
when using rolling = TRUE
whereby the incorrect number of periods were calulated. Thanks to Ganesh Krishnan for the bugfix.cvts()
function now allows additional arguments to be passed with ...
. Thanks to Ganesh Krishnan....
arguments can be passed to the individual component models in forecast.hybridModel()
.cvts()
function.forecast()
function from the "forecast" package when multiple or single prediction intervals are passed has changed. The prediction inervals are now consistently returned as matrices. This change fixes a bug in forecast.hybridModel()
when multiple prediction intervals are used.forecast.hybridModel()
for ets
, nnetar
, and stlm
component models when the level
argument was set to a single value instead of a vector of values.hybridModel()
nnetar
objects in the ensemble. This should address one aspect of incorrect prediction intervals (e.g. issue #37).f
" in the models =
argument for hybridModel()
) and are indeed part of the default - so by default, hybridModel() will now fit six modelsaccuracy.cvts()
is now exportedplot.hybridModel()
now supports ggplot2
graphics when the argument ggplot = TRUE
is passed.cvts()
weights = "cv.errors"
in hybridModel()
weights = "insample.errors"
and one or more component models perfectly fit the time seriesxreg
is included in n.args
but a nnetar
model is not included in the model listplot.hybridModel()
...
arguments to plot()
from plot.hybridModel()
print.hybridModel()
to three digits for cleaner displayverbose
argument and enable by default in hybridModel()
to display fitting/cross validation progressknitr rmarkdown
engineaccuracy()
and hybridModel.accuracy()
weights = "cv.errors"
nnetar
and stlm
models when 2 * frequency(y) >= length(y)
not()
function from "testthat" package2 * frequency(y) >= length(y)
, weights = "cv.errors"
)Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.