Panel to Plot Confidence Intervals by Groups in segplot()

This function allows non overlapping error bars in latticeExtra::segplot(). It has the groups = argument.

panel.groups.segplot(x, y, z, centers, groups, gap = NULL, data,
  subscripts, ...)

Arguments

x, y, z, centers, data, subscripts, ...	Arguments passed to segplot().
groups	The grouping variable (factor), with the same length of lwr e upr.
gap	Scalar that is the distance among segments. Default is 0.1. If the grouping factor has \(k\) levels, so \(0 \leq \textrm{gap} < 1/k\). A negative value for gap will put the segments in a reversed order.

See also

segplot().

Examples

library(latticeExtra)

m0 <- lm(log(breaks) ~ wool * tension, data = warpbreaks)

pred <- with(warpbreaks, expand.grid(KEEP.OUT.ATTRS = TRUE,
                                     wool = levels(wool),
                                     tension = levels(tension)))

pred <- cbind(pred,
              predict(m0, newdata = pred, interval = "confidence"))
str(pred)
#> 'data.frame':	6 obs. of  5 variables:
#>  $ wool   : Factor w/ 2 levels "A","B": 1 2 1 2 1 2
#>  $ tension: Factor w/ 3 levels "L","M","H": 1 1 2 2 3 3
#>  $ fit    : num  3.72 3.28 3.12 3.31 3.12 ...
#>  $ lwr    : num  3.47 3.03 2.87 3.06 2.87 ...
#>  $ upr    : num  3.97 3.53 3.37 3.56 3.37 ...

segplot(wool ~ lwr + upr, centers = fit, data = pred,
        draw = FALSE, horizontal = FALSE)

segplot(wool ~ lwr + upr, centers = fit, data = pred,
        draw = FALSE, horizontal = FALSE,
        groups = tension, gap = 0.05,
        panel = panel.groups.segplot)