Skip to contents

Automatically recalculate SPC control limits

create_SPC_auto_limits_table.Rd

create_SPC_auto_limits_table applies the Stable Shift Algorithm to automate recalculation of control limits.

Usage

create_SPC_auto_limits_table(
  data,
  chart_type,
  period_min,
  baseline_length,
  shift_rule_threshold,
  max_exclusions,
  no_regrets,
  verbosity,
  baseline_only,
  establish_every_shift,
  centre_line_tolerance,
  show_limits,
  overhanging_reversions,
  mr_screen_max_loops
)

Arguments

data

A data frame. For an XMR, C or C' chart, must have columns for:

  • the subgrouping variable, to be plotted on the horizontal axis, (x);

  • the variable of interest to be plotted on the vertical axis (y);

  • and optionally, a title and subtitle for the plot.


For a P or P' chart, must have columns for:

  • the subgrouping variable, to be plotted on the horizontal axis, (x);

  • the total count or denominator (n);

  • the count meeting criteria, or numerator (y);

  • and optionally, a title and subtitle for the plot.

chart_type

The type of chart you wish to plot. Available options are: "XMR", "MR", "C", "C'", "P", "P'".

Algorithm Parameters

Parameters that control behaviour of the algorithm used to re-establish control limits.

period_min

The minimum number of points (subgroups) per period, i.e. the minimum number of points required to form control limits.

baseline_length

Integer, overrides period_min for the first calculation period only, if specified

shift_rule_threshold

The minimum number of consecutive points above or below the centre line constituting a shift (or "rule 2") break.

max_exclusions

The maximum number of extreme points to exclude from limit calculations.

no_regrets

Boolean signifying which version of the algorithm should be used. Defines whether limits can change as more data is added or not.

verbosity

Integer 0-2 specifying how talkative the algorithm is in the standard output log; the higher the number the more information is provided, none if 0.

baseline_only

Boolean - if TRUE, do not recalculate control limits, instead extend limits calculated from the first period_min points.

establish_every_shift

Boolean - whether to bypass the Stable Shift Algorithm and simply re-establish limits at every shift rule break (respecting period_min)

centre_line_tolerance

Minimum difference between a point's vertical position and the centre line to count as "on the centre line" for the purposes ofshift rule breaks

show_limits

Boolean controlling whether or not to display centre line and control limits

overhanging_reversions

Boolean determining whether rule breaks in the opposite direction to a rule break triggering a candidate recalculation prevent recalculation even if they overhang the end of the candidate calculation period. Set to FALSE only with no_regrets = FALSE.

SPC Parameters

Parameters that control how cetnre line and control limits are established for each period, and details of how SPC rules are applied

mr_screen_max_loops

Integer or Inf specifying maximum number of times to recursively ignore mr values above the upper range limit when calculating xmr limits. Note this does not affect the calculation of the upper range limit on the mr chart.

Value

data frame with limits, rule breaks and additional info needed for plotting

Examples

# Calculate limts for a C' chart for count of monthly attendances

df <- ed_attendances_monthly %>%
        dplyr::rename(x = month_start,
                      y = att_all)

limits_table <- create_SPC_auto_limits_table(
  df,
  chart_type = "C'",
  period_min = 21,
  baseline_length = NULL,
  shift_rule_threshold = 8,
  max_exclusions = 3,
  no_regrets = TRUE,
  verbosity = 1L,
  baseline_only = FALSE,
  establish_every_shift = FALSE,
  centre_line_tolerance = 0,
  show_limits = TRUE,
  overhanging_reversions = TRUE,
  mr_screen_max_loops = 1L
)

head(limits_table)
#> # A tibble: 6 × 14
#>   x              y    ucl   lcl    cl periodType excluded breakPoint rule1 rule2
#>   <date>     <dbl>  <dbl> <dbl> <dbl> <chr>      <lgl>    <lgl>      <lgl> <lgl>
#> 1 2015-05-31 12178 10689. 8233. 9461. calculati… TRUE     NA         TRUE  FALSE
#> 2 2015-06-30 12888 10689. 8233. 9461. calculati… TRUE     FALSE      TRUE  FALSE
#> 3 2015-07-31 12360 10689. 8233. 9461. calculati… TRUE     FALSE      TRUE  FALSE
#> 4 2015-08-31 11232 10689. 8233. 9461. calculati… FALSE    FALSE      TRUE  FALSE
#> 5 2015-09-30 11445 10689. 8233. 9461. calculati… FALSE    FALSE      TRUE  FALSE
#> 6 2015-11-01  9330 10689. 8233. 9461. calculati… FALSE    FALSE      FALSE FALSE
#> # ℹ 4 more variables: aboveOrBelowCl <int>, highlight <chr>, runStart <lgl>,
#> #   log <chr>