inst/examples/ex-clt.R

In gWidgetsWWW2: Toolkit implementation of the gWidgets API for use with web pages

## Example of a CLT demonstration.
require(RSVGTipsDevice)

f <- get_tempfile(ext = ".svg")

makePlot <- function(h,...) {
  n <- min(as.numeric(svalue(sampleSize)), 500)
  m <- min(as.numeric(svalue(replicates)), 500)
  cmd <- pops[[svalue(population, index=TRUE)]]
  cmd$n <- n * m
  x <- matrix(do.call(cmd$cmd, cmd[-(1:2)]), nrow=n)
  xbars <- apply(x, 2, function(x) mean(x))
  svg(f)
  plot(density(xbars), main=sprintf("Population: %s, n: %s", svalue(population), svalue(sampleSize)))
  rug(xbars)
  dev.off()
  svalue(p) <- f
}

## Begin layout
w <- gwindow("CLT example")
gstatusbar("Powered by Rook and gWidgetsWWW2", cont = w)


bl <- gborderlayout(cont=w, horizontal=FALSE)

l <- glabel("Distribution of xbar, the sample mean.", cont = bl, where="north")

## Main widget is gsvg
width <- 480
height <- 400
p <- gsvg(f, width=width, height=height, cont = bl, where="center")
fg <- ggroup(horizontal=FALSE, cont=bl, where="west")
bl$set_panel_size("west", 300)

pops <- list("rnorm(n,0,1)" = list(cmd="rnorm", tooltip="Normal", mean=0, sd=1),
             "rexp(n)" = list(cmd="rexp", tooltip="Skewed"),
             "rt(n, df=3)" = list(cmd="rt",tooltip="Symmetric, long tailed", df=3),
             "rt(n, df=30)" = list(cmd="rt", tooltip="Symmetric, some tail", df=30),
             "runif(n)" = list(cmd="runif", tooltip="Short tail"),
             "rlnorm(n)" = list(cmd="rlnorm", tooltip="Very skewed")
             )
tbl <- glayout(cont = fg)
tbl[1,1] <- "n (sample size)"
tbl[1,2] <- (sampleSize <- gedit("10", cont = tbl, coerce.with=as.numeric))

tbl[2,1] <- "Population:"
tbl[2,2] <- (population <- gcombobox(names(pops), width="auto",
                                     editable=FALSE, cont=tbl))
tbl[3,1] <- "Replicates"
tbl[3,2] <- (replicates <- gslider(from=10, to=200, by=1, value=50, width=100,cont=tbl))
visible(tbl) <- TRUE

g1 <- ggroup(cont = fg)
b <- gbutton("plot", cont = g1, handler=makePlot)
helpButton <- gbutton("help", cont = g1, handler=function(h,...) {
  w1 <- gwindow("Help with CLT demo", width=400, height=300, parent=w)
  g <- ggroup(cont = w1, horizontal=FALSE, use.scrollwindow=TRUE)
  html <- "
<h1>CLT Example</h1>
Demonstrates the Central Limit Theorem, which says the CDF of the sample mean
of a random sample of size <i>n</i> converges to that of the normal distribution as,
<i>n</i> goes to infinity. This shows a density estimate, so it is not
quite the same for discrete, populations, but still illustrates the point.
<p>
<ul>
<li>Set the <b>sample size</b> by adjusting the value of <i>n</i>.</li>
<li>Set the <b>population</b> from one of the set ones.
The value <i>n</i>can be used as a parameter.</li>
<li>Adjust the number of <b>replicates</b> if desired.</li>
</ul>
Click the <b>plot</b> button to update the plot.
"
  ghtml(html, cont=g)
  gseparator(cont = g)
  bg <- ggroup(cont=g)
  gbutton("dismiss", cont = bg, handler=function(h,...) {
    dispose(w1)
  })
  visible(w1) <- TRUE
})

makePlot(1)