metRology-package {metRology} | R Documentation |
Support for Metrological Applications
Description
Provides classes and calculation and plotting functions for metrology applications, including measurement uncertainty estimation and inter-laboratory metrology comparison studies.
Details
The metRology package includes functions for:
Plotting for Key Comparisons (dot-and-bar, consistency)
Uncertainty evaluation using algebraic or numeric differentiation, with support for correlation
Monte Carlo evaluation of uncertainty (including correlation for normally distributed variables)
Classes and functions for location estimates for metrology comparisons
Mandel's h and k statistics and plots for interlaboratory studies
Support functions for an excel interface
Changes in version 0.9-29-2 from version 0.9-28-4 include:
A new set of functions for pair-difference chi-squared statistics; see
pdchisq
andbootPDchisq
.Group label placement in
blockplot
has been extended to takelegend
-like character specification.Variable name checking in
uncert
anduncertMC
has been harmonised and amended to fix a bug which caused apparent mismatches between the supplied variables inx
and the arguments in functionobj
when the function included an argument "...".Minor changes in help files and some functions to clear new CRAN checks.
WARNING: The addition of a new pairwise statistic involves significant duplication of code used for the bootMSD class. Future versions are likely to use common bootstrap code and classes for both MSD and PDchisq classes. bootMSD.default is accordingly deprecated and may be removed or reduced to a wrapper for boot.mtr.pairwise, the current workhorse for bootPDchisq.
Changes in version 0.9-28-4 from version 0.9-28-3:
Corrected bug in uncertMC which sometimes failed to include non-zero covariances in partial covariance treatment when negative.
Changes in version 0.9-28-3 from version 0.9-28-2 include:
Corrected huber.estimate to use additional arguments in rlm call.
Changes in version 0.9-28-2 from version 0.9-28-0 include:
Fixed an error occurring in
barplot.bootMSD
whennames.arg
was unspecified and the boot object had no labels.Fixed
blockplot
x-axis label, which was incorrect.
Changes in version 0.9-28-0 from version 0.9-27-2 include:
A new plot,
blockplot
, added. A “block plot” is a histogram variant identifiying individual data points, which appear as “blocks” in the plot.blockplot
provides for grouped data, which generates vertically separated subplots for each group. Fills and label colours can be specified for each data point.
Changes in version 0.9-27-2 from version 0.9-26-2 include:
-
pmsd
and related functions will now use fast interpolation by default, and provide exact values for both odd- and even-n
data sets up ton=199
. -
gplot
(called byplot.mandel.kh
) now has aspacing
parameter which allows finer control of vertical line spacing.
Changes in version 0.9-26-2 from version 0.9-26-1 include:
Fix to a bug in
reml.loc
which failed to report the standard uncertaintyu
correctly.-
cplot
now respectscex.axis
as a plot parameter.
Changes in version 0.9-26-1 from version 0.9-26-0 include:
Added plot and barplot methods for MSD class.
Minor correction to code in
msd
to prevent over-replication of estimateds
whens
is a function and returns a vector.
Changes in version 0.9-26 from version 0.9-25 include:
-
msd
now returns an object of class"MSD"
which includes the original data as attributes to permit subsequent bootstrapping. A new function,
bootMSD
that performs parametric bootstrapping forMSD
objects to obtain critical values and p-values for the general case where standard uncertainties/standard errors differ appreciably.
Improvements in version 0.9-25 from version 0.9-23 include:
-
plot.d.ellipse
now takes defaultxlab
andylab
from dimnames in the suppliedcov.dellipse
.
Improvements in version 0.9-23 from version 0.9-22 include:
A wholly new Youden plot (see
yplot
), with many options for confidence ellipsesA REML location estimate,
reml.loc
, in addition tovr.mle
;reml.loc
can use means and standard uncertainties/standard errors instead of raw data and when doing so does not require degrees of freedom.Incremental improvements in handling for the median scaled difference measure of anomalies.
msd
is faster and less memory-intensive, andpmsd
now uses a beta formulation to extend to very highn
(at least 1e6 - if you feel very patient).Support for
log
andlog.p
indt.scaled
.
Corrections and bugfixes include:
amends
plot.drop1.uncert
to give a plot for each measure of change specified inwhich
corrects a grep warning appearing in
drop1.uncert
;corrects an unnecessary 'missing
u
' error message in version 0.9-22'suncert()
whencov
was specified andu
was not.
Author(s)
Stephen L R Ellison. [aut], Stephen L R Ellison [cre]
Maintainer: Stephen L R Ellison <s.ellison@lgcgroup.com>