Released tinyM5 data set, a subset of the M5
dataset.
Included NNARMA (non-negative autoregressive moving
average) model by Sbrana & Babai (2026).
Added occurrence smoothing for TWEES to determine
the dispersion parameter.
Added Tweedie distribution as an option in the parametric distribution model.
Renamed EMPDISTR and PARAMDISTR to
EMPSD and PARAMSD.
Dropped mixture of distribution in PARAMSD.
Fixed a bug in the information criteria in
PARAMSD.
Changed optimisation strategy for damped exponential smoothing models.
Fixed recursion in BETANBB and
GAMPOISB.
Implemented non-negative Gaussian forecast distribution for ARMA-based models.
Increased test coverage.
pasta hierarchical data set.dist_tweedie() and stats-like
Tweedie functions; moved them to standalone tweedieDistr
package.Added report() and tidy() methods for
all model classes.
Details included in model_sum() for ES-based models
and PARAMSD().
Fixed an error in the Tweedie quantile method
(quantile.dist_tweedie()).
MARWAL() forecast distributions are now truncated at
zero.