qcc object and
optional within group standard deviation or control limit specifications.
Rules may be applied to the chart and violating points highlighted.
shewhart(object, newdata)
shewhart(object, newdata, type, limits, target, std.dev, sizes,
labels, label.limits = c("LCL","UCL"),
confidence.level = 0.999, nsigmas, add.stats = T,
chart.all = F, ylim.min = NULL, rules = shewhart.rules,
highlight = list(pch = 1, col = 2), ...)
"qcc" which provides information on the type of group
summary statistics to plot and the within group standard deviation necessary
for computing the control limits.
statistics,
new.statistics, or both in
object are charted depending
on the argument
chart.all.
"xbar",
"s",
"R",
"p",
"np",
"u", or
"c" built-in functions will be used. (See
help(stats.xbar) for more information on these functions.) If the name
of a function, it is used to compute the group summary statistics.
std.dev component in
object
or the
std.dev argument if it is given. If missing, the
limits function
determined by
type is used to compute the control limits. See DETAILS below.
center component of
object.
std.dev
component of
object.
newdata for xbar, s,
and R charts. For p, np, and u charts,
sizes corresponds to the number of
units examined. For c charts, sizes is assumed to be 1.
sizes may be used in conjunction with providing summary statistics
rather than the original new data to
newdata and is necessary if type is
"p",
"np" or
"u". If of length 1, its value is assumed to be the size of
all the groups.
newdata. The default vector
of labels is the
"names" attribute if
newdata is a vector or list, the
first component of the
"dimnames" attribute if
newdata is a matrix, the
"row.names" attribute if
newdata is a data frame, or, if the appropriate
attribute is NULL, the first n integers where n is the number of groups.
c("LCL","UCL").
confidence.level is used.
TRUE. Set
add.stats = FALSE if you want to
add your own text or lines to the chart.
statistics component of
object
should be plotted along with the
new.statistics component of
object if
present and the summary statistics of
newdata if given. The default of
FALSE does not plot the
statistics component of
object.
shewhart.rules.
See DETAILS below.
rules. The default is
highlight = list(pch = 1, col = 2). If more than
one rule is applied to the chart, this can be a list of lists of plotting
parameters for highlighting successively the points violating each rule.
rules.
"shewhart" object, which inherits from
"qcc", with all
the components of
object plus the following additional arguments:
newdata with "names" attribute
equal to labels.
If this exists in
object prior to calling
shewhart and
newdata is given,
the group summary statistics computed from
newdata will be appended to the
new.statistics
component of
object and the entire set of statistics
returned.
new.sizes component of
object exists prior to calling
shewhart and
newdata
is given,
new.sizes computed from
newdata will be appended to the
new.sizes
component of
object and the entire set of sizes returned.
target argument if specified.
newdata.
If this exists in
object prior to calling
shewhart and
newdata is given,
an
"and" and the name of the input data will be appended to the
newdata.name
component of
object and the entire character string returned.
shewhart can pass the return value of a call to
shewhart
, which includes
newdata, in place of
object with additional
data in the
newdata argument.
The argument
object provides the chart type (see
help(qcc) for more details)
and an estimate for the within group standard deviation if
std.dev is not
given as an argument. The control limits are computed and a Shewhart chart is
produced for three different sets of summary statistics depending on the
value of
chart.all and whether
newdata is given.
1) If
chart.all = F and
newdata
is given, any summary statistics in the
new.statistics component
of
object and the summary statistics for
newdata computed depending
upon
type will be charted.
2) If
chart.all = F and
newdata is missing, only the statistics in the
new.statistics
component of
object will be charted.
3) If
chart.all = T,
the
statistics component and the
new.statistics component of
object will
be charted along with the summary statistics of
newdata if given.
If
limits is missing,
shewhart will use the
cntrl.limits component
of
object, if it exists, as the control limits for the chart. If
limits
is missing and there is not a
cntrl.limits component of
object, appropriate
default functions compute the control limits for the chart. The name of the
default function that computes the control limits is formed by pasting together
"limits"
, a period, and the value of
type. Thus the
control limits of an xbar chart are computed by
.
By default, probability limits are computed with equal tail probabilities
computed as
(1 - confidence.level)/2. The control limits are centered in
probability about
target if it is given, the
target component of
object
if present, or the
center component of
object, in that order or preference.
If
nsigmas is given,
in the normal case
the control limits are computed as
target (or
center
if
target and
object$target are missing) plus or minus
nsigmas times the
standard errors of the group summary statistics.
In the binomial case a variance-stabilizing transformation is
performed first, limits are computed, and results translated back to
the original scale; see
.
If
length(unique(sizes)) > 1,
the summary statistics' standard errors will be
different, and the control limits are computed as step functions.
For more details on how the control limits are computed for Shewhart
charts see the help file for the appropriate function (e.g.
for
the xbar chart).
If
limits is given as a vector or matrix, the values specify the control
limits.
If
limits is a function, then it must return a vector or matrix
specifying the control limits. If it returns a vector, the first element
is the lower control limit and the second element is the upper control
limit. If it returns a matrix, the first column corresponds to the lower control
limits and the second column corresponds to the upper control limits.
The
ylim.min argument can be used to expand the range of the y-axis if
additional
detail needs to be added to the chart. For example, additional limits can be
added to a chart by setting
ylim.min appropriately. If the additional limits
are in
addl.limits, then setting
ylim.min = addl.limits will guarantee that
the y-axis range is large enough for any subsequent calls to
abline or
lines
. When low level plotting functions are used to add detail to a
Shewhart chart, you need to specify
add.stats = FALSE in the call to
shewhart
or else the coordinates will be incorrect for adding detail.
The
rules argument specifies a function to determine points which violate
a given (set of) rule(s). For example, the
beyond.limits function returns
the indices of the charted statistics which are beyond the control limits. The
indicated points are highlighted according the plotting parameters specified
by
highlight. Other rules functions which are provided are
runs.target for
determining all points in a run, for a given run length, above or below the
target
value and
shewhart.rules which combines
beyond.limits and
runs.target
. Although
rules can be used in this way to more quickly
detect small shifts (on the order of 0.5 - 1.5 standard errors) in the
process, cusum charts are preferable. See
help(cusum) for more details.
std.dev argument, or the
std.dev
component of
object in that order of preference.
If
newdata is given, the summary statistics for
newdata along with
any summary statistics contained in the
new.statistics component of
object
are charted. If
newdata is missing, the
new.statistics component of
object
is charted. If
chart.all = TRUE the
statistics component of
object
is included on the chart. Repeated calls to
shewhart with the return
value of the previous call as
object and a
newdata argument each time
charts the summary statistics of the aggregate of the
newdata arguments.
limits.ewma
,
limits.mR
,
limits.ma
, and
limits.ms
.
shewhart(qcc(matrix(rnorm(100), ncol = 5), type = "xbar"))