Changelog

The plot.nppi() function now returns an additional plot for JAB Pseudo Value Stability.

In nppi() there was an inconsistency in dimensionality between the jab_point_values and the jab_pseudo_values due to how NA values were handled. This is fixed such that both vectors will always have the same length.

The nppi() function now returns an additional item in the nppi object: jab_pseudo_values which include the ith MBJ pseudo-value used in the JAB variance estimator.

In nppi() there was an incorrect computation of the JAB variance estimator, which did use the ith MBJ pseudo-value in place of the the ith MBJ point value. This is now fixed.

A new function nppi() has been added to the package. This function estimates the optimal block-length using the non-parametric plug-in (NPPI) method of Lahiri, Furukawa, Lee, (2007).
A new S3 plot method plot.nppi() has been added to the package. This function outputs a diagnostic plot for JAB point values calculated from the above nppi() function.

In pwsd(), m_hat was indexing the first insignificant lag of the correlation structure rather than the first significant lag prior to the consecutive run of insignificant lags.
pwsd() and plot.pwsd() now output significance bands that match rho_k_critical exactly. Prior to this, the significance bands on the correlogram output were generated using the plot.acf(ci = ) argument which led to misleading graphical representations of the implied hypothesis test’s critical value in the PWSD method.
In hhj(), sub_block_length has been changed to sub_sample to avoid confusion with the other tuning parameter, pilot_block_length

A new vignette has been included on tuning and diagnosing problematic output from the selection functions!
pwsd() now includes a new argument to override the implied hypothesis test by setting m_hat directly.
In pwsd(), rho.k.critical has been changed to snake case rho_k_critical in the $parameters matrix from output of class ‘pwsd’ objects from pwsd().
In hhj(), if subsample = is set directly it is now rounded to the nearest whole number.
plot.pwsd() now includes “darkmagenta” significance lines to match pwsd().
plot.pwsd() now explicitly includes an option to customize title with main = .
hhj() now includes a warning message if the supplied iteration limit n_iter is reached but still outputs an object of class ‘hhj.’

blocklength (development version)