deseats 1.1.2
create.gain() and runDecomposition() are
now deprecated and replaced by create_gain() and
run_decomposition().
select_bwidth(), deseats() and
s_semiarma() now have an argument
inf_criterion that lets the user switch between
"bic" (the default) and "aic"; this selects
the information criterion (BIC or AIC) to use for the ARMA order
selection wherever relevant within those functions.
select_bwidth(), deseats() and
s_semiarma() now warn the user, when an extremely small
number of observations is being used, which makes the bandwidth
selection highly unreliable.
- some internal functions have been made more robust, so that they do
not throw an error under small numbers of observations; instead they
warn the user (see previous point) and fall back to certain minimum /
maximum settings that still work (but are unreliable).
- some defaults in
set_option() have been adjusted,
namely order_poly = 3 instead of
order_poly = 1 as well as
boundary_method = "shorten" instead of
boundary_method = "extend".
- a vignette was added to the package that introduces readers to the
basic applications of the package with an example.
- the
animate() method for output of
deseats() now also shows proper results for the case with
boundary_method = "shorten" in
set_options().
- the
predict() method for S-Semi-ARMA models now allows
for bias adjustment of point forecasts via
bias.adjust = TRUE, if expo = TRUE.
deseats 1.1.1
- fixed a bug in the forecasting of seasonal semiparametric ARMA
models, where for the special case of an ARMA for the rest including a
mean estimate the forecasting intervals under the normality assumption
where too large by exactly that mean estimate
- fixed an issue, probably due to some updates in
ggplot2, where in autoplotfor forecasting
objects the legend for multiple forecasting intervals was not displayed
correctly anymore
deseats 1.1.0
- the argument
correction_factor in the functions
deseats, s_semiarma and
select_bwidth is now set to FALSE by
default.
- further arguments were added to
deseats,
s_semiarma and select_bwidth to allow for a
fine-adjustments of the algorithm, when an ARMA model is assumed as the
error process.
- the functions
seasonplot and seasonplot_gg
were added to allow for the creation of seasonal plots.
- the faulty link in the manual to the
RcppArmadillo
package was removed.
- the example in the documentation of
ts_conversion was
adjusted to use file.path for the path to the temporary
data file.