createQtlPlot: Create the QTL plot
In PuzzledFace/qtlr: An Implementation of Quality Tolerance Limits

Description Usage Arguments Value Usage Notes Examples

Create the QTL plot

createQtlPlot(mcmcData, mcmcVar = Value, groupData,
  groupResponse = ObservedResponse, groupSize = N, groupType = NULL,
  basicTheme = ggplot2::theme_light, qtls = c(0.1, 0.2, 0.8, 0.9),
  postScaleFactor = NULL, mcmcAlpha = 0.2, nDensity = 512,
  xAxisRange = NULL)

`mcmcData`	the tibble containing the posterior MCMC samples of the metric
`mcmcVar`	the column in `mcmcData` containing the sampled values of the metric
`groupData`	the tibble containing the observed group data
`groupResponse`	the column in `groupData` containing the observed by-group metric values
`groupSize`	the column in the `groupData` containing the group sizes
`groupType`	The column in `groupData` containing the group type for for the observation. If `NULL`, group types are not identified in the plot. Otherwise, `groupType` is converted to a factor and used as an aesthetic.
`basicTheme`	one of the themes provided by `ggplot2`, for example `theme_light` or `theme_dark`.
`qtls`	a list containing the quantiles that define the QTL thresholds. If `NULL` or of zero length, the QTL thresholds are not shown.
`postScaleFactor`	scales the posterior density relative to the height of the `groupSize` bars. If `NULL`, `postScaleFactor` is chosen so that the mode of the posterior is twice as high as the tallest group size bar. See Usage Notes below.
`mcmcAlpha`	the transparency of the fill used to indicate QTL bands
`nDensity`	the number of bins used to estimate the density. Must be a power of two. Normally, this can be left unchanged, but occasionally a different value is needed to correct visual artefacts related to the shading of the plot
`xAxisRange`	the range of the x-axis, as defined by a call to `coord_cartesian()`. NULL uses the default axis range.

the QTL plot

The plot returned by the function is deliberately crude: the x axis is unlabelled, the legends are unfomatted, and so on. However, it can be easily customised to include more appropriate titles, legends and so forth, as shown in the second example.

The scaled kernel estimate of the density is multiplied by postScaleFactor to obtain the curve shown in the plot. The height of the scaled kernel density at its mode is 1. Therefore, by default, postScaleFactor is set to twice the largest group size.

b <- createBerryData() %>% 
       tibble::add_column(Region=as.factor(c(1,1,1,2,2,2,3,3,3)))
m <- fitBinomialModel(b$Subjects, b$Events) %>% dplyr::filter(Index==10)
createQtlPlot(groupData=b,
              groupResponse=ObservedResponse,
              groupSize=Subjects,
              groupType=Region,
              mcmcData=m,
              mcmcVar=Value) +
              ggplot2::scale_fill_manual(name="QTL",
                                         labels=c("Action low",
                                                  "Warn low",
                                                  "",
                                                  "Warn high",
                                                  "Action high"),
                                         values=c("red",
                                                  "yellow",
                                                  "white",
                                                  "yellow",
                                                  "red")) +
              ggplot2::scale_colour_manual(name="Region",
                                           values=1:3,
                                           labels=c("US",
                                                    "EMEA",
                                                    "Other")) +
              ggplot2::labs(x="Observed response",
                            title="Example QTL plot")