#' Elemental Graphic for Display of Dependent Sample Data
#'
#' Plots dependent sample data beginning from a scatterplot for the X,Y pairs;
#' proceeds to display difference scores as point projections; also X and Y
#' means, as well as the mean of the difference scores.
#'
#' Paired X and Y values are plotted as scatterplot. The identity reference line
#' (for Y = X) is drawn. Parallel projections of data points to (a lower-left)
#' line segment show how each point relates to its X-Y = D difference;
#' semitransparent "shadow" points are used to display the distribution of
#' difference scores, with thin grey lines leading from each raw datapoint to
#' its shadow projection on the difference distribution. The range of that
#' difference score distribution is drawn as a blue line beneath the shadow
#' points and the mean difference is displayed as a heavy dashed purple line,
#' parallel to the identity reference line. Means for X and Y are also plotted
#' (as thin dashed vertical and horizontal lines), and rug plots are shown for
#' the distributions of X (at the top of graphic) and Y (on the right side). The
#' 95\% confidence interval for the population mean difference is also shown
#' graphically as a green band, perpendicular to the mean treatment effect line.
#' Because all data points are plotted relative to the identity line, and
#' summary results are shown graphically, clusters, data trends, outliers, and
#' possible uses of transformations are readily seen, possibly to be
#' accommodated.
#'
#' In summary, the graphic shows all initial data points relative to the
#' identity line, adds projections (to the 'north' and 'east') showing the
#' marginal distributions of X and Y, as well as projections to the 'southwest'
#' where the difference scores for each point are drawn. Means for all three
#' distributions are shown using straight lines; the confidence interval for the
#' population mean difference score is also shown. Summary statistics are
#' printed as side effects of running the function for the dependent sample
#' analysis.
#'
#' @param data is an n X 2 dataframe or matrix. First column defines X
#' (intially for horzontal axis), the second defines Y.
#' @param main optional main title (as character); can be supplied by user. The default value is
#' \code{"default_granova_title"}, which will print a generic title for the graphic.
#' @param revc reverses X,Y specifications
#' @param xlab optional label (as character) for horizontal axis. If not
#' defined, axis labels are taken from colnames of data.
#' @param ylab optional label (as character) for vertical axis. If not
#' defined, axis labels are taken from colnames of data.
#' @param conf.level The confidence level at which to perform a dependent sample t-test.
#' Defaults to \code{0.95} (95\% Confidence)
#' @param plot.theme argument indicating a ggplot2 theme to apply to the
#' graphic; defaults to a customized theme created for the dependent sample graphic
#' @param northeast.padding (numeric) extends axes toward lower left,
#' effectively moving data points to the southwest. Defaults to zero padding.
#' @param southwest.padding (numeric) extends axes toward upper right,
#' effectively moving data points to the southwest. Defaults to zero padding.
#' Making both southwest and northeast padding smaller moves points farther apart,
#' while making both larger moves data points closer together.
#' @param ... Optional arguments to/from other functions
#' @return Returns a plot object of class \code{ggplot}.
#'
#' @author Brian A. Danielak \email{brian@@briandk.com}\cr
#' Robert M. Pruzek \email{RMPruzek@@yahoo.com}
#'
#' with contributions by:\cr
#' William E. J. Doane \email{wil@@drdoane.com}\cr
#' James E. Helmreich \email{James.Helmreich@@Marist.edu}\cr
#' Jason Bryer \email{jason@@bryer.org}
#'
#' @seealso \code{\link{granovagg.1w}},
#' \code{\link{granovagg.ds}}, \code{\link{granovaGG}}
#'
#' @example demo/granovagg.ds.R
#' @export
#' @references Pruzek, R. M., & Helmreich, J. E. (2009). Enhancing Dependent Sample Analyses with Graphics. Journal of Statistics Education, 17(1), 21.
#' @references Wickham, H. (2009). Ggplot2: Elegant Graphics for Data Analysis. New York: Springer.
#' @references Wilkinson, L. (1999). The Grammar of Graphics. Statistics and computing. New York: Springer.
granovagg.ds2 <- function(input,
data = NULL,
revc = FALSE,
main = "default_granova_title",
xlab = NULL,
ylab = NULL,
conf.level = 0.95,
plot.theme = "theme_granova_ds",
northeast.padding = 0,
southwest.padding = 0,
...
)
{
GetData <- function(data) {
data <- CheckData(data)
data <- ReverseXAndY(data)
data <- EnsureDataHasColumnNames(data)
data <- EnsureDataIsADataFrame(data)
return(data)
}
CheckData <- function(data) {
IsDataNull(data)
IsDataInTwoColumnFormat(data)
return(data)
}
ReverseXAndY <- function(data) {
output <- data
if (revc) {
output[, 1] <- data[, 2]
output[, 2] <- data[, 1]
names(output) <- names(data)[2:1]
}
return(output)
}
IsDataNull <- function(data) {
if (is.null(length(data))) {
stop("It looks like you didn't pass any data to granovagg.ds")
}
}
IsDataInTwoColumnFormat <- function(data) {
message <- "It looks like the data you handed in isn't in two-column (n x 2) format. granovagg.ds needs n x 2 data to work."
if (is.null(dim(data))) {
stop(message)
}
if (dim(data)[2] != 2) {
stop(message)
}
}
EnsureDataIsADataFrame <- function(data) {
output <- data
if (!is.data.frame(data)) {
output <- as.data.frame(output)
}
return(output)
}
EnsureDataHasColumnNames <- function(data) {
output <- data
if (is.null(colnames(data))) {
colnames(data) <- c("x", "y")
}
return(data)
}
GetXs <- function(data) {
return(data[, 1])
}
GetYs <- function(data) {
return(data[, 2])
}
GetEffect <- function(dsp) {
return(GetXs(dsp$data) - GetYs(dsp$data))
}
GetTtest <- function(data, conf.level) {
return(t.test(data[, 1],
data[, 2],
paired = TRUE,
conf.level = conf.level
)
)
}
GetStats <- function(dsp, conf.level) {
ttest <- GetTtest(dsp$data, conf.level)
return(data.frame(lower.treatment.effect = as.numeric(ttest$conf.int[2]),
mean.treatment.effect = as.numeric(ttest$estimate[1]),
upper.treatment.effect = as.numeric(ttest$conf.int[1]),
t.statistic = as.numeric(ttest$statistic[1])
)
)
}
GetGraphicsParams <- function(dsp) {
.aggregate.data.range <- c(range(GetXs(dsp$data)), range(GetYs(dsp$data)))
.extrema <- c(max(.aggregate.data.range), min(.aggregate.data.range))
.square.data.range <- max(.extrema) - min(.extrema)
.southwest.padding <- (65/100) * .square.data.range
.northeast.padding <- (15/100) * .square.data.range
.lower.graphical.bound <- min(.extrema) - .southwest.padding
.upper.graphical.bound <- max(.extrema) + .northeast.padding
.bounds <- c(.lower.graphical.bound, .upper.graphical.bound)
.center <- mean(.bounds)
.crossbow.anchor <- mean(.bounds) + min(.bounds)
.shadow.offset <- (1/100)*.square.data.range
return(list(square.data.range = .square.data.range,
bounds = .bounds,
shadow.offset = .shadow.offset,
anchor = .crossbow.anchor,
point.size = I(4),
mean.line.size = I(1/2)
)
)
}
GetShadows <- function(dsp) {
x.shadow <- (dsp$effect / 2) +
(3 * dsp$params$bounds[1] + dsp$params$bounds[2]) / 4 +
(4 * dsp$params$shadow.offset)
y.shadow <- x.shadow - dsp$effect
return(data.frame(x.shadow, y.shadow))
}
GetCrossbow <- function(dsp) {
return(data.frame(x = min(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y = max(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset),
x.end = max(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y.end = min(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset)
)
)
}
GetCIBand <- function(dsp) {
return(data.frame(x.end = ((dsp$params$anchor + dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y.end = ((dsp$params$anchor - dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
x = ((dsp$params$anchor + dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y = ((dsp$params$anchor - dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
color = factor(paste(100 * conf.level, "% CI", " (t = ", round(dsp$stats$t.statistic, digits = 2), ")", sep =""))
)
)
}
GetTreatmentLine <- function(dsp) {
return(data.frame(intercept = -dsp$stats$mean.treatment.effect,
slope = 1,
color = factor(paste("Mean Diff. =", round(dsp$stats$mean.treatment.effect, digits = 2)))
)
)
}
GetCrossElementCoordinates <- function(dsp) {
color <- NULL # to appease R CMD check
crossbow <- dsp$crossbow
ci.band <- subset(dsp$CIBand, select = -color)
output <- rbind(crossbow, ci.band)
return(output)
}
EnsureCrossElementsAppearInVisualBounds <- function(dsp) {
minimum <- min(dsp$params$bounds,
dsp$cross.elements$x,
dsp$cross.elements$y.end
)
return(c(minimum - dsp$params$shadow.offset, max(dsp$params$bounds)))
}
GetTrails <- function(dsp) {
return(data.frame(x.trail.start = GetXs(dsp$data),
y.trail.start = GetYs(dsp$data),
x.trail.end = GetXs(dsp$shadow),
y.trail.end = GetYs(dsp$shadow)
)
)
}
GetColors <- function(dsp) {
return(list(treatment.line = "#542570",
rugplot = "black",
mean.line = "red",
CIBand = "#33A02C",
crossbow = "#377EB8",
datapointedge = "black",
datapointfill = "grey80",
diff = "dodgerblue"
)
)
}
PrintSummary <- function(dsp) {
summary <- GetPrintedSummary(dsp)
summary <- RenamePrintedSummaryRows(summary)
summary <- round(summary, digits = 3)
print(summary)
}
GetPrintedSummary <- function(dsp) {
n <- dim(dsp$data)[1]
mean.1 <- mean(dsp$data[, 1])
mean.2 <- mean(dsp$data[, 2])
mean.d <- mean.1 - mean.2
standard.deviation.d <- sd(dsp$data[, 1] - dsp$data[, 2])
effect.size <- (mean.d / standard.deviation.d)
r.xy <- cor(dsp$data[, 1], dsp$data[, 2])
r.x.plus.y.d <- cor((dsp$data[, 1] + dsp$data[, 2]), (dsp$data[, 1] - dsp$data[, 2]))
lower.treatment.confidence <- dsp$stats$upper.treatment.effect
upper.treatment.confidence <- dsp$stats$lower.treatment.effect
t.value <- dsp$t.test$statistic
degrees.of.freedom <- dsp$t.test$parameter
p.value <- dsp$t.test$p.value
return(matrix(c(n,
mean.1,
mean.2,
mean.d,
standard.deviation.d,
effect.size,
r.xy,
r.x.plus.y.d,
lower.treatment.confidence,
upper.treatment.confidence,
t.value,
degrees.of.freedom,
p.value
),
ncol = 1
)
)
}
RenamePrintedSummaryRows <- function(summary) {
dimnames(summary) <- list(
c("n",
paste(colnames(dsp$data)[1], "mean"),
paste(colnames(dsp$data)[2], "mean"),
paste("mean(D = ", colnames(dsp$data)[1], " - ", colnames(dsp$data)[2], ")", sep = ""),
paste("SD(D)"),
paste("Effect Size"),
paste("r(", colnames(dsp$data)[1], ", ", colnames(dsp$data)[2], ")", sep = ""),
paste("r(", colnames(dsp$data)[1], " + ", colnames(dsp$data)[2], ", D)", sep = ""),
paste("Lower ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("Upper ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("t (D-bar)"),
paste("df.t"),
paste("p-value (t-statistic)")
), "Summary Statistics")
return(summary)
}
dsp <- list(data = GetData(data))
dsp$effect <- GetEffect(dsp)
dsp$stats <- GetStats(dsp, conf.level)
dsp$t.test <- GetTtest(dsp$data, conf.level)
dsp$params <- GetGraphicsParams(dsp)
dsp$shadows <- GetShadows(dsp)
dsp$crossbow <- GetCrossbow(dsp)
dsp$CIBand <- GetCIBand(dsp)
dsp$cross.elements <- GetCrossElementCoordinates(dsp)
dsp$params$bounds <- EnsureCrossElementsAppearInVisualBounds(dsp)
dsp$treatment.line <- GetTreatmentLine(dsp)
dsp$trails <- GetTrails(dsp)
dsp$colors <- GetColors(dsp)
#PrintSummary(dsp)
# Because of the way ggplot2 creates plot objects, layers can be
# added to a plot p simply by calling "p <- p + newLayer"
InitializeGgplot <- function(dsp) {
return(
ggplot(
aes_string(x = names(dsp$data)[1],
y = names(dsp$data)[2]
), data = dsp$data
)
)
}
TreatmentLine <- function(dsp) {
return(
geom_abline(
aes_string(
intercept = "intercept",
slope = "slope",
color = "color"
),
alpha = 0.5,
size = I(1),
linetype = "dashed",
data = dsp$treatment.line
)
)
}
RawData1 <- function(dsp) {
return(
geom_point(
size = dsp$params$point.size,
color= dsp$colors$datapointedge)
)
}
RawData2 <- function(dsp) {
return(
geom_point(
size = dsp$params$point.size-1,
color= dsp$colors$datapointfill)
)
}
IdentityLine <- function() {
return(
geom_abline(
slope = 1,
intercept = 0,
alpha = 0.75,
size = 1
)
)
}
ScaleX <- function(dsp) {
return(
scale_x_continuous(
limits = dsp$params$bounds
)
)
}
ScaleY <- function(dsp) {
return(
scale_y_continuous(
limits = dsp$params$bounds
)
)
}
PadViewingWindow <- function(params) {
ne.offset = params$square.data.range * southwest.padding
sw.offset = params$square.data.range * northeast.padding
padded.window = c(params$bounds[1] - sw.offset, params$bounds[2] + ne.offset)
return(
coord_cartesian(
xlim = padded.window,
ylim = padded.window
)
)
}
RugPlot <- function(dsp) {
return(
geom_rug(
size = 1/2,
alpha = 1/3,
color = dsp$colors$rugplot,
sides = "tr", # top and right sides
data = dsp$data
)
)
}
XMeanLine <- function(dsp) {
return(
geom_vline(
xintercept = mean(GetXs(dsp$data)),
color = dsp$colors$mean.line,
size = dsp$params$mean.line.size,
linetype = "dashed",
alpha = I(1/2)
)
)
}
YMeanLine <- function(dsp) {
return(
geom_hline(
yintercept = mean(GetYs(dsp$data)),
color = dsp$colors$mean.line,
size = dsp$params$mean.line.size,
linetype = "dashed",
alpha = I(1/2)
)
)
}
Crossbow <- function(dsp) {
return(
geom_segment(
aes_string(x = "x",
y = "y",
xend = "x.end",
yend = "y.end"
),
size = 3/4,
alpha = 3/4,
color = dsp$colors$crossbow,
data = dsp$crossbow
)
)
}
CIBand <- function(dsp) {
return(
geom_segment(
aes_string(
x = "x",
y = "y",
xend = "x.end",
yend = "y.end",
color = "color"
),
size = 2,
data = dsp$CIBand
)
)
}
Shadows <- function(dsp) {
return(
geom_point(
aes_string(
x = "x.shadow",
y = "y.shadow"
),
data = dsp$shadow,
size = dsp$params$point.size,
shape = 16,
alpha = 1/2,
color=dsp$colors$diff
)
)
}
Trails <- function(dsp) {
return(
geom_segment(
aes_string(
x = "x.trail.start",
y = "y.trail.start",
xend = "x.trail.end",
yend = "y.trail.end"
),
data = dsp$trails,
size = 1/3,
color = "black",
linetype = 1,
alpha = 1/10
)
)
}
ColorScale <- function(dsp) {
colors <- c(dsp$colors$treatment.line, dsp$colors$CIBand)
return(scale_color_manual(values = colors, name = ""))
}
XLabel <- function(dsp) {
result <- colnames(dsp$data)[1]
if(!is.null(xlab)) {
result <- xlab
}
return(xlab(result))
}
YLabel <- function(dsp) {
result <- colnames(dsp$data)[2]
if(!is.null(ylab)) {
result <- ylab
}
return(ylab(result))
}
Title <- function(main) {
output.title <- "Dependent Sample Assessment Plot"
if (main != "default_granova_title") {
output.title <- main
}
return(
ggtitle(output.title)
)
}
ForceCoordinateAxesToBeEqual <- function() {
return(coord_fixed(ratio = 1))
}
if(input$extent==1) {
p <- InitializeGgplot(dsp)
p <- p + RawData1(dsp)+ RawData2(dsp)
p <- p + Theme(plot.theme)
p <- p + ScaleX(dsp) + ScaleY(dsp)
p <- p + PadViewingWindow(dsp$params)
p <- p + ForceCoordinateAxesToBeEqual()
p <- p + Title(main)
p <- p + XLabel(dsp)
p <- p + YLabel(dsp)
p <- p + RugPlot(dsp)
p <- p + XMeanLine(dsp) + YMeanLine(dsp)
p <- p + TreatmentLine(dsp)
p <- p + IdentityLine()
p <- p + Shadows(dsp)
p <- p + Crossbow(dsp)
p <- p + Trails(dsp)
p <- p + ColorScale(dsp)
p <- p + CIBand(dsp)
return(p)
}
else if(input$extent==2){
p <- InitializeGgplot(dsp)
p <- p + RawData1(dsp)+ RawData2(dsp)
p <- p + Theme(plot.theme)
p <- p + ScaleX(dsp) + ScaleY(dsp)
p <- p + PadViewingWindow(dsp$params)
p <- p + ForceCoordinateAxesToBeEqual()
p <- p + Title(main)
p <- p + XLabel(dsp)
p <- p + YLabel(dsp)
if(input$rug){
p <- p + RugPlot(dsp)}
if(input$means){
p <- p + XMeanLine(dsp) + YMeanLine(dsp)}
if(input$unit){
p <- p + IdentityLine()}
if(input$diffs){
p <- p + Shadows(dsp)
p <- p + Crossbow(dsp)}
if(input$trails){
p <- p + Trails(dsp)}
if(input$meandiff){
p <- p + TreatmentLine(dsp)}
if(input$ci){
p <- p + CIBand(dsp)
p <- p + ColorScale(dsp)}
return(p)
}
}
ds.summary<- function(input,
data = NULL,
revc = FALSE,
main = "default_granova_title",
xlab = NULL,
ylab = NULL,
conf.level = 0.95,
plot.theme = "theme_granova_ds",
northeast.padding = 0,
southwest.padding = 0,
...
)
{
GetData <- function(data) {
data <- CheckData(data)
data <- ReverseXAndY(data)
data <- EnsureDataHasColumnNames(data)
data <- EnsureDataIsADataFrame(data)
return(data)
}
CheckData <- function(data) {
IsDataNull(data)
IsDataInTwoColumnFormat(data)
return(data)
}
ReverseXAndY <- function(data) {
output <- data
if (revc) {
output[, 1] <- data[, 2]
output[, 2] <- data[, 1]
names(output) <- names(data)[2:1]
}
return(output)
}
IsDataNull <- function(data) {
if (is.null(length(data))) {
stop("It looks like you didn't pass any data to granovagg.ds")
}
}
IsDataInTwoColumnFormat <- function(data) {
message <- "It looks like the data you handed in isn't in two-column (n x 2) format. granovagg.ds needs n x 2 data to work."
if (is.null(dim(data))) {
stop(message)
}
if (dim(data)[2] != 2) {
stop(message)
}
}
EnsureDataIsADataFrame <- function(data) {
output <- data
if (!is.data.frame(data)) {
output <- as.data.frame(output)
}
return(output)
}
EnsureDataHasColumnNames <- function(data) {
output <- data
if (is.null(colnames(data))) {
colnames(data) <- c("x", "y")
}
return(data)
}
GetXs <- function(data) {
return(data[, 1])
}
GetYs <- function(data) {
return(data[, 2])
}
GetEffect <- function(dsp) {
return(GetXs(dsp$data) - GetYs(dsp$data))
}
GetTtest <- function(data, conf.level) {
return(t.test(data[, 1],
data[, 2],
paired = TRUE,
conf.level = conf.level
)
)
}
GetStats <- function(dsp, conf.level) {
ttest <- GetTtest(dsp$data, conf.level)
return(data.frame(lower.treatment.effect = as.numeric(ttest$conf.int[2]),
mean.treatment.effect = as.numeric(ttest$estimate[1]),
upper.treatment.effect = as.numeric(ttest$conf.int[1]),
t.statistic = as.numeric(ttest$statistic[1])
)
)
}
GetGraphicsParams <- function(dsp) {
.aggregate.data.range <- c(range(GetXs(dsp$data)), range(GetYs(dsp$data)))
.extrema <- c(max(.aggregate.data.range), min(.aggregate.data.range))
.square.data.range <- max(.extrema) - min(.extrema)
.southwest.padding <- (65/100) * .square.data.range
.northeast.padding <- (15/100) * .square.data.range
.lower.graphical.bound <- min(.extrema) - .southwest.padding
.upper.graphical.bound <- max(.extrema) + .northeast.padding
.bounds <- c(.lower.graphical.bound, .upper.graphical.bound)
.center <- mean(.bounds)
.crossbow.anchor <- mean(.bounds) + min(.bounds)
.shadow.offset <- (1/100)*.square.data.range
return(list(square.data.range = .square.data.range,
bounds = .bounds,
shadow.offset = .shadow.offset,
anchor = .crossbow.anchor,
point.size = I(4),
mean.line.size = I(1/2)
)
)
}
GetShadows <- function(dsp) {
x.shadow <- (dsp$effect / 2) +
(3 * dsp$params$bounds[1] + dsp$params$bounds[2]) / 4 +
(4 * dsp$params$shadow.offset)
y.shadow <- x.shadow - dsp$effect
return(data.frame(x.shadow, y.shadow))
}
GetCrossbow <- function(dsp) {
return(data.frame(x = min(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y = max(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset),
x.end = max(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y.end = min(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset)
)
)
}
GetCIBand <- function(dsp) {
return(data.frame(x.end = ((dsp$params$anchor + dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y.end = ((dsp$params$anchor - dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
x = ((dsp$params$anchor + dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y = ((dsp$params$anchor - dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
color = factor(paste(100 * conf.level, "% CI", " (t = ", round(dsp$stats$t.statistic, digits = 2), ")", sep =""))
)
)
}
GetTreatmentLine <- function(dsp) {
return(data.frame(intercept = -dsp$stats$mean.treatment.effect,
slope = 1,
color = factor(paste("Mean Diff. =", round(dsp$stats$mean.treatment.effect, digits = 2)))
)
)
}
GetCrossElementCoordinates <- function(dsp) {
color <- NULL # to appease R CMD check
crossbow <- dsp$crossbow
ci.band <- subset(dsp$CIBand, select = -color)
output <- rbind(crossbow, ci.band)
return(output)
}
EnsureCrossElementsAppearInVisualBounds <- function(dsp) {
minimum <- min(dsp$params$bounds,
dsp$cross.elements$x,
dsp$cross.elements$y.end
)
return(c(minimum - dsp$params$shadow.offset, max(dsp$params$bounds)))
}
GetTrails <- function(dsp) {
return(data.frame(x.trail.start = GetXs(dsp$data),
y.trail.start = GetYs(dsp$data),
x.trail.end = GetXs(dsp$shadow),
y.trail.end = GetYs(dsp$shadow)
)
)
}
GetColors <- function(dsp) {
return(list(treatment.line = "#542570",
rugplot = "black",
mean.line = "red",
CIBand = "#33A02C",
crossbow = "#377EB8",
datapointedge = "black",
datapointfill = "grey80",
diff = "dodgerblue"
)
)
}
PrintSummary <- function(dsp) {
summary <- GetPrintedSummary(dsp)
summary <- RenamePrintedSummaryRows(summary)
summary <- round(summary, digits = 3)
print(summary)
}
GetPrintedSummary <- function(dsp) {
n <- dim(dsp$data)[1]
mean.1 <- mean(dsp$data[, 1])
mean.2 <- mean(dsp$data[, 2])
mean.d <- mean.1 - mean.2
standard.deviation.d <- sd(dsp$data[, 1] - dsp$data[, 2])
effect.size <- (mean.d / standard.deviation.d)
r.xy <- cor(dsp$data[, 1], dsp$data[, 2])
r.x.plus.y.d <- cor((dsp$data[, 1] + dsp$data[, 2]), (dsp$data[, 1] - dsp$data[, 2]))
lower.treatment.confidence <- dsp$stats$upper.treatment.effect
upper.treatment.confidence <- dsp$stats$lower.treatment.effect
t.value <- dsp$t.test$statistic
degrees.of.freedom <- dsp$t.test$parameter
p.value <- dsp$t.test$p.value
return(matrix(c(n,
mean.1,
mean.2,
mean.d,
standard.deviation.d,
effect.size,
r.xy,
r.x.plus.y.d,
lower.treatment.confidence,
upper.treatment.confidence,
t.value,
degrees.of.freedom,
p.value
),
ncol = 1
)
)
}
RenamePrintedSummaryRows <- function(summary) {
dimnames(summary) <- list(
c("n",
paste(colnames(dsp$data)[1], "mean"),
paste(colnames(dsp$data)[2], "mean"),
paste("mean(D = ", colnames(dsp$data)[1], " - ", colnames(dsp$data)[2], ")", sep = ""),
paste("SD(D)"),
paste("Effect Size"),
paste("r(", colnames(dsp$data)[1], ", ", colnames(dsp$data)[2], ")", sep = ""),
paste("r(", colnames(dsp$data)[1], " + ", colnames(dsp$data)[2], ", D)", sep = ""),
paste("Lower ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("Upper ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("t (D-bar)"),
paste("df.t"),
paste("p-value (t-statistic)")
), "Summary Statistics")
return(summary)
}
dsp <- list(data = GetData(data))
dsp$effect <- GetEffect(dsp)
dsp$stats <- GetStats(dsp, conf.level)
dsp$t.test <- GetTtest(dsp$data, conf.level)
dsp$params <- GetGraphicsParams(dsp)
dsp$shadows <- GetShadows(dsp)
dsp$crossbow <- GetCrossbow(dsp)
dsp$CIBand <- GetCIBand(dsp)
dsp$cross.elements <- GetCrossElementCoordinates(dsp)
dsp$params$bounds <- EnsureCrossElementsAppearInVisualBounds(dsp)
dsp$treatment.line <- GetTreatmentLine(dsp)
dsp$trails <- GetTrails(dsp)
dsp$colors <- GetColors(dsp)
PrintSummary(dsp)
}
require(shiny)
library(ggplot2)
#library(Cairo)
load("anorexia.sub.rda")
# Define server logic required to generate and plot a random distribution
shinyServer(function(input, output){
source('tools.R')
# source('tools2.R')
source('granovaGG.ds.R')
#
#
makeplot <- function(input){
if(input$plottype==1){
p <- granovagg.ds2(input,
data=anorexia.sub,
revc = FALSE,
main = "Assessment Plot for Body Weights Before \
and After Family Therapy in Anorexia Patients",
xlab = "Weight before therapy (lbs.)",
ylab = "Weight after therapy (lbs.)"
)
}
else if(input$plottype==2){
p <- ggplot(anorexia.sub,aes(x=anorexia.sub$Postwt-anorexia.sub$Prewt, y=..density..))
p <- p + geom_histogram(fill="cornsilk",binwidth=4, colour="grey65",size=.2)
p <- p + geom_line(stat="density")+xlim(-10,25)
#
#hist(anorexia.sub$Postwt-anorexia.sub$Prewt,prob=T,
# main="Pre and Post-treatment Body Weight Differences",
# xlab="Difference Scores")
#lines(density(anorexia.sub$Postwt-anorexia.sub$Prewt,col="skyblue"))
}
return(p)
}
output$about <- renderRmd('about.Rmd', input)
output$aboutdata <- renderRmd('aboutdata.Rmd', input)
output$distPlot <- renderPlot({
makeplot(input)
},width=600)
output$table1 <- renderTable({
ds.summary(data=anorexia.sub,recv=TRUE)
}
)
output$slides <- renderText({
if (input$docs==1){
return(paste0('<iframe style="height:600px; width:100%"',
'src = "jsepaper.pdf"></iframe>'))
}
else if (input$docs==2){
return(paste0('<iframe style="height:600px; width:100%"',
'src = "granova.pdf"></iframe>'))
}
else if (input$docs==3){
return(paste0('<iframe style="height:600px; width:100%"',
'src = "granovaGG.pdf"></iframe>'))
}
}) # finish outputslides
output$test <- renderRmd('testfile.Rmd', input)
output$downloadPNG <- downloadHandler(
filename = function() { paste0("test1.png") },
content = function(file1) {
png(file1)
makeplot(input)
dev.off()
}) #finish downloadpng
output$downloadPDF <- downloadHandler(
filename = function() { paste0("test1.pdf") },
content = function(file1) {
CairoPDF(file1)
makeplot(input)
dev.off()
if (file.exists(paste0(file1, ".pdf")))
file.rename(paste0(file1, ".pdf"), file1)
}) # finish downloadpdf
})# finish shinyserver
load("anorexia.sub.rda")
ds.summary<- function(input,
data = NULL,
revc = FALSE,
main = "default_granova_title",
xlab = NULL,
ylab = NULL,
conf.level = 0.95,
plot.theme = "theme_granova_ds",
northeast.padding = 0,
southwest.padding = 0,
...
)
{
GetData <- function(data) {
data <- CheckData(data)
data <- ReverseXAndY(data)
data <- EnsureDataHasColumnNames(data)
data <- EnsureDataIsADataFrame(data)
return(data)
}
CheckData <- function(data) {
IsDataNull(data)
IsDataInTwoColumnFormat(data)
return(data)
}
ReverseXAndY <- function(data) {
output <- data
if (revc) {
output[, 1] <- data[, 2]
output[, 2] <- data[, 1]
names(output) <- names(data)[2:1]
}
return(output)
}
IsDataNull <- function(data) {
if (is.null(length(data))) {
stop("It looks like you didn't pass any data to granovagg.ds")
}
}
IsDataInTwoColumnFormat <- function(data) {
message <- "It looks like the data you handed in isn't in two-column (n x 2) format. granovagg.ds needs n x 2 data to work."
if (is.null(dim(data))) {
stop(message)
}
if (dim(data)[2] != 2) {
stop(message)
}
}
EnsureDataIsADataFrame <- function(data) {
output <- data
if (!is.data.frame(data)) {
output <- as.data.frame(output)
}
return(output)
}
EnsureDataHasColumnNames <- function(data) {
output <- data
if (is.null(colnames(data))) {
colnames(data) <- c("x", "y")
}
return(data)
}
GetXs <- function(data) {
return(data[, 1])
}
GetYs <- function(data) {
return(data[, 2])
}
GetEffect <- function(dsp) {
return(GetXs(dsp$data) - GetYs(dsp$data))
}
GetTtest <- function(data, conf.level) {
return(t.test(data[, 1],
data[, 2],
paired = TRUE,
conf.level = conf.level
)
)
}
GetStats <- function(dsp, conf.level) {
ttest <- GetTtest(dsp$data, conf.level)
return(data.frame(lower.treatment.effect = as.numeric(ttest$conf.int[2]),
mean.treatment.effect = as.numeric(ttest$estimate[1]),
upper.treatment.effect = as.numeric(ttest$conf.int[1]),
t.statistic = as.numeric(ttest$statistic[1])
)
)
}
GetGraphicsParams <- function(dsp) {
.aggregate.data.range <- c(range(GetXs(dsp$data)), range(GetYs(dsp$data)))
.extrema <- c(max(.aggregate.data.range), min(.aggregate.data.range))
.square.data.range <- max(.extrema) - min(.extrema)
.southwest.padding <- (65/100) * .square.data.range
.northeast.padding <- (15/100) * .square.data.range
.lower.graphical.bound <- min(.extrema) - .southwest.padding
.upper.graphical.bound <- max(.extrema) + .northeast.padding
.bounds <- c(.lower.graphical.bound, .upper.graphical.bound)
.center <- mean(.bounds)
.crossbow.anchor <- mean(.bounds) + min(.bounds)
.shadow.offset <- (1/100)*.square.data.range
return(list(square.data.range = .square.data.range,
bounds = .bounds,
shadow.offset = .shadow.offset,
anchor = .crossbow.anchor,
point.size = I(4),
mean.line.size = I(1/2)
)
)
}
GetShadows <- function(dsp) {
x.shadow <- (dsp$effect / 2) +
(3 * dsp$params$bounds[1] + dsp$params$bounds[2]) / 4 +
(4 * dsp$params$shadow.offset)
y.shadow <- x.shadow - dsp$effect
return(data.frame(x.shadow, y.shadow))
}
GetCrossbow <- function(dsp) {
return(data.frame(x = min(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y = max(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset),
x.end = max(dsp$shadows$x.shadow) - (2 * dsp$params$shadow.offset),
y.end = min(dsp$shadows$y.shadow) - (2 * dsp$params$shadow.offset)
)
)
}
GetCIBand <- function(dsp) {
return(data.frame(x.end = ((dsp$params$anchor + dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y.end = ((dsp$params$anchor - dsp$stats$lower.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
x = ((dsp$params$anchor + dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
y = ((dsp$params$anchor - dsp$stats$upper.treatment.effect) / 2) - 3 * (dsp$params$shadow.offset),
color = factor(paste(100 * conf.level, "% CI", " (t = ", round(dsp$stats$t.statistic, digits = 2), ")", sep =""))
)
)
}
GetTreatmentLine <- function(dsp) {
return(data.frame(intercept = -dsp$stats$mean.treatment.effect,
slope = 1,
color = factor(paste("Mean Diff. =", round(dsp$stats$mean.treatment.effect, digits = 2)))
)
)
}
GetCrossElementCoordinates <- function(dsp) {
color <- NULL # to appease R CMD check
crossbow <- dsp$crossbow
ci.band <- subset(dsp$CIBand, select = -color)
output <- rbind(crossbow, ci.band)
return(output)
}
EnsureCrossElementsAppearInVisualBounds <- function(dsp) {
minimum <- min(dsp$params$bounds,
dsp$cross.elements$x,
dsp$cross.elements$y.end
)
return(c(minimum - dsp$params$shadow.offset, max(dsp$params$bounds)))
}
GetTrails <- function(dsp) {
return(data.frame(x.trail.start = GetXs(dsp$data),
y.trail.start = GetYs(dsp$data),
x.trail.end = GetXs(dsp$shadow),
y.trail.end = GetYs(dsp$shadow)
)
)
}
GetColors <- function(dsp) {
return(list(treatment.line = "#542570",
rugplot = "black",
mean.line = "red",
CIBand = "#33A02C",
crossbow = "#377EB8",
datapointedge = "black",
datapointfill = "grey80",
diff = "dodgerblue"
)
)
}
PrintSummary <- function(dsp) {
summary <- GetPrintedSummary(dsp)
summary <- RenamePrintedSummaryRows(summary)
summary <- round(summary, digits = 3)
print(summary)
}
GetPrintedSummary <- function(dsp) {
n <- dim(dsp$data)[1]
mean.1 <- mean(dsp$data[, 1])
mean.2 <- mean(dsp$data[, 2])
mean.d <- mean.1 - mean.2
standard.deviation.d <- sd(dsp$data[, 1] - dsp$data[, 2])
effect.size <- (mean.d / standard.deviation.d)
r.xy <- cor(dsp$data[, 1], dsp$data[, 2])
r.x.plus.y.d <- cor((dsp$data[, 1] + dsp$data[, 2]), (dsp$data[, 1] - dsp$data[, 2]))
lower.treatment.confidence <- dsp$stats$upper.treatment.effect
upper.treatment.confidence <- dsp$stats$lower.treatment.effect
t.value <- dsp$t.test$statistic
degrees.of.freedom <- dsp$t.test$parameter
p.value <- dsp$t.test$p.value
return(matrix(c(n,
mean.1,
mean.2,
mean.d,
standard.deviation.d,
effect.size,
r.xy,
r.x.plus.y.d,
lower.treatment.confidence,
upper.treatment.confidence,
t.value,
degrees.of.freedom,
p.value
),
ncol = 1
)
)
}
RenamePrintedSummaryRows <- function(summary) {
dimnames(summary) <- list(
c("n",
paste(colnames(dsp$data)[1], "mean"),
paste(colnames(dsp$data)[2], "mean"),
paste("mean(D = ", colnames(dsp$data)[1], " - ", colnames(dsp$data)[2], ")", sep = ""),
paste("SD(D)"),
paste("Effect Size"),
paste("r(", colnames(dsp$data)[1], ", ", colnames(dsp$data)[2], ")", sep = ""),
paste("r(", colnames(dsp$data)[1], " + ", colnames(dsp$data)[2], ", D)", sep = ""),
paste("Lower ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("Upper ", (100 * conf.level), "% ", "Confidence Interval", sep = ""),
paste("t (D-bar)"),
paste("df.t"),
paste("p-value (t-statistic)")
), "Summary Statistics")
return(summary)
}
dsp <- list(data = GetData(data))
dsp$effect <- GetEffect(dsp)
dsp$stats <- GetStats(dsp, conf.level)
dsp$t.test <- GetTtest(dsp$data, conf.level)
dsp$params <- GetGraphicsParams(dsp)
dsp$shadows <- GetShadows(dsp)
dsp$crossbow <- GetCrossbow(dsp)
dsp$CIBand <- GetCIBand(dsp)
dsp$cross.elements <- GetCrossElementCoordinates(dsp)
dsp$params$bounds <- EnsureCrossElementsAppearInVisualBounds(dsp)
dsp$treatment.line <- GetTreatmentLine(dsp)
dsp$trails <- GetTrails(dsp)
dsp$colors <- GetColors(dsp)
PrintSummary(dsp)
}
ds.summary(data=anorexia.sub,revc=T)
#win.graph()
#granovagg.ds(anorexia.sub,
# revc = TRUE,
# main = "Assessment Plot for weights to assess\
# Family Therapy treatment for Anorexia Patients",
# xlab = "Weight after therapy (lbs.)",
# ylab = "Weight before therapy (lbs.)"
#)
#library(granovaGG)
library(grid)
library(ggplot2)
load("anorexia.sub.rda")
# function to render markdown text for about panel and other things
renderRmd <- function(path, input){
return(renderText( {
if (!require(knitr))
stop("knitr package is not installed")
if (!require(markdown))
stop("Markdown package is not installed")
# shiny:::dependsOnFile(path)
contents <- paste(readLines(path, warn = FALSE), collapse = '\n')
myenvir <- new.env() # Perhaps use parent.frame() ?!
assign('input', input, envir=myenvir)
html <- knitr::knit2html(text = contents, fragment.only = TRUE, envir=myenvir)
Encoding(html) <- 'UTF-8'
return(HTML(html))
}))
}
theme_granova_ds <- function(base_size = 12) {
theme_granova_ds <- theme_bw()
theme_granova_ds$axis.title.x <- element_text(size = 12)
theme_granova_ds$axis.title.y <- element_text(size = 12, angle = 90)
theme_granova_ds$legend.key.size <- unit(1, "lines")
theme_granova_ds$legend.text <- element_text(size = 8, lineheight = 8)
theme_granova_ds$plot.title <- element_text(face = "bold", size = base_size)
return(theme_granova_ds)
}
########################
# shared functions
LayoutFourPlotsPerPage <- function(list.of.plots) {
four.plot.pages <- floor(length(list.of.plots) / 4)
remainder.plots <- length(list.of.plots) %% 4 # a %% b is notation for a modulo b
LayoutFourPlotPages(list.of.plots, four.plot.pages)
LayoutRemainderPlots(list.of.plots, remainder.plots)
}
LayoutFourPlotPages <- function(plot.list, pages) {
i <- 1
while (i < 4 * pages) {
args.list <- list(plot.list[[i]],
plot.list[[i+1]],
plot.list[[i+2]],
plot.list[[i+3]]
)
args.list <- c(args.list, list(nrow = 2, ncol = 2))
do.call(grid.arrange, args.list)
DisplayEndOfPageMessage()
i <- i + 4
}
}
LayoutRemainderPlots <- function(plot.list, remainder.plots) {
if (remainder.plots > 0) {
remainder.start <- length(plot.list) - remainder.plots + 1
remainder.end <- length(plot.list)
args.list <- lapply(X = remainder.start : remainder.end,
FUN = GetListElementByIndex,
x = plot.list
)
args.list <- c(args.list, list(nrow = 2, ncol = 2))
do.call(grid.arrange, args.list)
}
}
GetListElementByIndex <- function(x, index) {
return(x[[index]])
}
DisplayEndOfPageMessage <- function() {
if (interactive()) {
readline("Examine the contrast plots and consider printing. When you're done, press <Return>")
}
}
Theme <- function(plot.theme) {
return(
eval(call(plot.theme))
)
}
GetContrastName <- function(contrast.data, index) {
if (is.null(dimnames(contrast.data))) {
contrast.name <- paste(index)
}
else {
provided.contrast.name <- dimnames(contrast.data)[[2]][index]
contrast.name <- paste(provided.contrast.name)
}
return(paste("Contrast ", contrast.name, sep=""))
}
# ReorderDataByColumn reorders a data structure according to the values in "column"
# For example, "sort the data on all these cars according to their highway mileage"
ReorderDataByColumn <- function(x, column) {
return(x[order(x[[column]]), ])
}
OverlapWarning <- function(data, tolerance) {
first <- FirstElementOverlaps(data, tolerance)
inner <- InnerElementsOverlap(data, tolerance)
last <- LastElementOverlaps(data, tolerance)
return(c(first, inner, last))
}
FirstElementOverlaps <- function(data, tolerance) {
return(
abs(data[1] - data[2]) < tolerance
)
}
InnerElementsOverlap <- function(data, tolerance) {
return(sapply(
X = 2:(length(data) - 1),
FUN = NeighborOverlaps,
data = data,
tolerance = tolerance
)
)
}
LastElementOverlaps <- function(data, tolerance) {
last.index <- length(data)
return(
abs(data[last.index] - data[(last.index) - 1]) < tolerance
)
}
NeighborOverlaps <- function(data, tolerance, index) {
left.overlap <- abs(data[index] - data[(index - 1)]) < tolerance
right.overlap <- abs(data[index] - data[(index + 1)]) < tolerance
return(left.overlap | right.overlap)
}
# Define UI for application that plots random distributions
shinyUI(pageWithSidebar(
# Application title
headerPanel("Dependent Samples Illustration of GRANOVA.DS -- UNDER CONSTRUCTION"),
# Sidebar with controls
sidebarPanel(
conditionalPanel(condition="input.conditionedPanels==1",
HTML("Choose Tabs for display of the graphical App and Other Docs")
),
conditionalPanel(condition="input.conditionedPanels==2",
selectInput(
"dataset", "Choose Data Set to Evaluate: (only anorexia data set now)",
c("Family Therapy for Anorexia"="anorexia",
"Tobacco Leaf Lesions"="tobacco",
"Shoe Wear"="shoes",
"Surgical Stress and Endorphin Levels"="endorphin",
"Matched Pairs and Blood Levels"="lead",
"Upload your own data"="userdata")
),
radioButtons("plottype", label=h4("Choose Type of Plot"),
choices=list("Granova Dependent Samples Plot" = 1,
"Frequency Histogram of Difference Scores"=2),
selected=1),
hr(),
conditionalPanel(condition="input.plottype==1",
radioButtons("extent", label=h4("Choose approach to granova.ds Graph"),
choices=list("Full Dependent Samples Plot"=1,
"Build Plot by Adding Elements"=2),
selected=1),
conditionalPanel("input.extent==2",
wellPanel(
HTML("Start with Base ScatterPlotand then add:"),
checkboxInput("rug", label=h5("Univariate Rug Plots"), value=FALSE),
checkboxInput("means", label=h5("Means of X and Y"), value=FALSE),
checkboxInput("unit", label=h5("Line with Unit Slope"), value=FALSE),
checkboxInput("diffs", label=h5("Difference Scores"), value=FALSE),
checkboxInput("trails", label=h5("Connect Difference scores to XY Score"), value=FALSE),
checkboxInput("meandiff", label=h5("Mean of Difference Scores"), value=FALSE),
checkboxInput("ci", label=h5("Confidence Interval for Difference Scores"), value=FALSE)
#checkboxInput("rug", label=h5("Add Univariate Rug Plots"), value=FALSE)
) # end wellpanel
) # end conditionalpanel for extent=2
),#end conditionalpanel for plottype1
conditionalPanel("input.plottype==2",
wellPanel(
))# ends conditionalpanel for plottype2
), # ends conditionedPanel for plots
conditionalPanel(condition="input.conditionedPanels==3"),
conditionalPanel(condition="input.conditionedPanels==4",
radioButtons("docs", label=h4("Choose Document for view"),
choices=list("Pruzek and Helmreich (2009) JSE article"=1,
"granova package documentation"=2,
"granovaGG package documentation"=3),
selected=1)
),#ends conditionalpanel 4
conditionalPanel(condition="input.conditionedPanels==5")
)# ends sidebarpanel
,
# Show a plot of the generated distribution
mainPanel(
tabsetPanel(
tabPanel("About This App", value=1,
htmlOutput('about')),
tabPanel("Plot", value=2,
plotOutput("distPlot"),
tableOutput("table1")),
tabPanel("About the Data",value=3,
htmlOutput('aboutdata')),
tabPanel("Documents",value=4,
htmlOutput('slides')),
tabPanel("tester", value=5,
htmlOutput('test')
)
,
id="conditionedPanels"
)
)
))