Is there a package in R that can do a variability plot? A variability plot is a kind of categorized dot plot. (If there is a lot of data in each category, box plots are used rather than dot plots.) Usually, the categories are factor level combinations. All the dot plots appear in the same window; below the x-axis a hierarchy of factors shows which dot plot corresponds to which factor-level combination. Examples can be seen http://statsoft.com/support/blog/entryid/64/user-defined-variability-plots/ and http://www.public.iastate.edu/~wrstephe/stat495/GaugeRR_WaferThickness_JMPOutput.pdf By reordering the factor names in the function call, the user can reorder the factor level combinations on the graph, making it easier to do the visual comparisons of interest. The user should also have the option to draw line segments at factor level combination means/medians, and to connect the category means/medians to make visual comparison easier. The only softwares which I am aware of which produce such a plot are Statistica and JMP. I have found these plots to be more powerful than lattice-style categorizations in their ability to allow the user to conveniently process experimental data visually. [[alternative HTML version deleted]]
On May 20, 2011, at 7:12 PM, Joseph Boyer wrote:> Is there a package in R that can do a variability plot? > > A variability plot is a kind of categorized dot plot. (If there is a > lot of data in each category, box plots are used rather than dot > plots.) > Usually, the categories are factor level combinations. All the dot > plots appear in the same window; below the x-axis a hierarchy of > factors > shows which dot plot corresponds to which factor-level combination. > > Examples can be seen > http://statsoft.com/support/blog/entryid/64/user-defined-variability-plots/ > and > http://www.public.iastate.edu/~wrstephe/stat495/GaugeRR_WaferThickness_JMPOutput.pdfI'm afraid I developed an allergic response to SixSigma DMAIC bul....t after being dragged through a GreenBelt indoctrination session. The "teachers" did not want to hear about either regression or nonparametric statistical approaches because it wasn't in their approved curriculum. I'm sure you can whip up something quite like that with R. In fact it would be good teaching exercise. Go for it! You could alternatively read the Posting Guide about what is requested of a poster on R-help.> > By reordering the factor names in the function call, the user can > reorder the factor level combinations on the graph, making it easier > to do the visual comparisons of interest. The user should also have > the option to draw line segments at factor level combination means/ > medians, and to connect the category means/medians to make visual > comparison easier. > > The only softwares which I am aware of which produce such a plot are > Statistica and JMP. I have found these plots to be more powerful than > lattice-style categorizations in their ability to allow the user to > conveniently process experimental data visually.David Winsemius, MD West Hartford, CT
Here's one attempt; I only used five of the wafers since you didn't
provide any data.
dd <- data.frame(wafer = factor(rep(1:5, each = 6)),
operator = factor(rep(rep(1:3, each = 2), 5)),
thickness = c(0.62, 0.66, 0.53, 0.53, 0.51, 0.55,
0.99, 1.00, 1.05, 0.93, 1.05, 1.02,
0.82, 0.81, 0.80, 0.77, 0.90, 0.77,
0.85, 0.89, 0.83, 0.76, 0.79, 0.81,
0.59, 0.48, 0.39, 0.40, 0.46, 0.51))
# Summarize the data to output the mean, sd, min and max of thickness
library(ggplot2)
dsumm <- ddply(dd, .(wafer, operator), summarise, tmean = mean(thickness),
tmin = min(thickness), tmax = max(thickness),
tsd = sd(thickness))
# 'Multi-vari' plot:
p1 <- ggplot(dd) +
geom_point(aes(x = wafer, y = thickness)) +
geom_errorbar(data = dsumm, aes(x = wafer, y = tmean,
ymin = tmin, ymax = tmax), colour = 'blue') +
geom_segment(data = dsumm, aes(x = wafer, y = tmean, yend = tmean,
xend = as.numeric(wafer) + 0.2), colour = 'blue') +
geom_segment(data = dsumm, aes(x = wafer, y = tmean, yend = tmean,
xend = as.numeric(wafer) - 0.2), colour = 'blue') +
facet_wrap( ~ operator, nrow = 1) + xlab("")
# Standard deviation plot
p2 <- ggplot(dsumm, aes(x = wafer, y = tsd)) +
geom_point(colour = 'blue') + geom_line(aes(group = 1), size 1,
colour = 'blue') +
facet_wrap( ~ operator, nrow = 1)
# Use the gridExtra package to combine the two graphs
library(gridExtra)
grid.arrange(p1, p2)
HTH,
Dennis
On Fri, May 20, 2011 at 4:12 PM, Joseph Boyer <joseph.g.boyer at gsk.com>
wrote:> Is there a package in R that can do a variability plot?
>
> A variability plot is a kind of categorized dot plot. (If there is a lot of
data in each category, box plots are used rather than dot plots.)
> Usually, the categories are factor level combinations. All the dot plots
appear in the same window; below the x-axis a hierarchy of factors
> shows which dot plot corresponds to which factor-level combination.
>
> Examples can be seen
> http://statsoft.com/support/blog/entryid/64/user-defined-variability-plots/
> and
>
http://www.public.iastate.edu/~wrstephe/stat495/GaugeRR_WaferThickness_JMPOutput.pdf
>
> By reordering the factor names in the function call, the user can reorder
the factor level combinations on the graph, making it easier
> to do the visual comparisons of interest. The user should also have the
option to draw line segments at factor ?level combination means/medians, and to
connect the category means/medians to make visual comparison easier.
>
> The only softwares which I am aware of which produce such a plot are
Statistica and JMP. I have found these plots to be more powerful than
> lattice-style categorizations in their ability to allow the user to
conveniently process experimental data visually.
>
> ? ? ? ?[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>