- The
plot.nppi()function now returns an additional plot for JAB Pseudo Value Stability.
- In
nppi()there was an inconsistency in dimensionality between thejab_point_valuesand thejab_pseudo_valuesdue 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 thenppiobject:jab_pseudo_valueswhich 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 abovenppi()function.
- Minor adjustments to documentation per CRAN requests.
-
In
pwsd(),m_hatwas indexing the first insignificant lag of the correlation structure rather than the first significant lag prior to the consecutive run of insignificant lags. -
pwsd()andplot.pwsd()now output significance bands that matchrho_k_criticalexactly. Prior to this, the significance bands on the correlogram output were generated using theplot.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_lengthhas been changed tosub_sampleto 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 settingm_hatdirectly. -
In
pwsd(),rho.k.criticalhas been changed to snake caserho_k_criticalin the$parametersmatrix from output of class 'pwsd' objects frompwsd(). -
In
hhj(), ifsubsample =is set directly it is now rounded to the nearest whole number. -
plot.pwsd()now includes "darkmagenta" significance lines to matchpwsd(). -
plot.pwsd()now explicitly includes an option to customize title withmain = . -
hhj()now includes a warning message if the supplied iteration limitn_iteris reached but still outputs an object of class 'hhj.'
- This is the first submission accepted to CRAN
- Minor changes to documentation and testing of examples
- This is the first submission of blocklength package.