Graphical Display 2

Graphical Display 2

Pictures of Data

Line Graphs

• At least two variables, x is often time or depth while the other(s) are plotted for comparison

• Cumulative graphs allow comparison of two or more distributions (Quantile comparison plot)

> Nelson Depth Corrugated Biscuit Type_I Type_II_Red Type_II_Yellow Type_II_Gray Type_III1 1 57 10 2 24 23 34 52 2 116 17 2 64 90 76 63 3 27 2 10 68 18 48 34 4 28 4 6 52 20 21 05 5 60 15 2 128 55 85 06 6 75 21 8 192 53 52 17 7 53 10 40 91 20 15 08 8 56 2 118 45 1 5 09 9 93 1 107 3 0 0 010 10 84 1 69 0 0 0 0> NelsonPct <- data.frame(Nelson[,1:2], prop.table(as.matrix(Nelson[,3:8]),1)*100)> NelsonPct Depth Corrugated Biscuit Type_I Type_II_Red Type_II_Yellow Type_II_Gray Type_III1 1 57 10.2040816 2.0408163 24.489796 23.4693878 34.693878 5.10204082 2 116 6.6666667 0.7843137 25.098039 35.2941176 29.803922 2.35294123 3 27 1.3422819 6.7114094 45.637584 12.0805369 32.214765 2.01342284 4 28 3.8834951 5.8252427 50.485437 19.4174757 20.388350 0.00000005 5 60 5.2631579 0.7017544 44.912281 19.2982456 29.824561 0.00000006 6 75 6.4220183 2.4464832 58.715596 16.2079511 15.902141 0.30581047 7 53 5.6818182 22.7272727 51.704545 11.3636364 8.522727 0.00000008 8 56 1.1695906 69.0058480 26.315789 0.5847953 2.923977 0.00000009 9 93 0.9009009 96.3963964 2.702703 0.0000000 0.000000 0.000000010 10 84 1.4285714 98.5714286 0.000000 0.0000000 0.000000 0.0000000

# Vital Stats - Crude Birth Rate, Crude Death Rate, Expectancy of# Life at Birth", Infant Mortality, and Total Fertility# Rate from 2000 to 2010top5 <- c("China", "India", "United States", "Indonesia", "Brazil")Top5<-VitalStats[VitalStats$Country %in% top5, c(substr(colnames(VitalStats), 1, 3) %in% c("IMR", "Cou"))]rownames(Top5) <- Top5$CountryTop5$Country <- NULLTop5IMR <- t(Top5)Year <- as.numeric(substr(rownames(Top5IMR), 4, 7))Top5IMR <- data.frame(Year, Top5IMR)rownames(Top5IMR) <- 1:9Top5IMR Year United.States Brazil China India Indonesia1 2000 7.0 35.2 30.3 54.9 40.92 2001 7.0 34.0 28.9 51.5 39.53 2002 6.9 32.9 27.7 48.2 38.24 2003 6.8 31.7 26.4 45.2 36.95 2004 6.6 30.7 25.3 42.4 36.46 2005 6.5 29.6 24.2 39.7 34.57 2006 6.4 28.6 23.1 37.1 33.38 2007 6.4 27.6 22.1 34.6 32.19 2010 6.2 24.9 19.4 28.1 28.9

Quantile comparison plot

Scatterplot – XY Plot

• Two interval/ratio variables• Smoothed and regression lines,

linear and non-linear relationships• Compare groups (ellipses)• Label points (outliers)• Scatterplot matrix to compare

more than two variables

Scatterplot

• Numerous options. Turn off all options on the menus, but select “Plot by groups” and select Name

• Insert three options into the command:– legend.coords=“topleft”– ellipse=TRUE– levels=.95

Scatterplot Matix

• This gives you a visual display of a correlation matrix between three or more variables

• Default puts a kernel density plot in the diagonal with a rug showing the data points

• For values: by(DartPoints[,6:8], DartPoints$Name, rcorr.adjust)

> library(Rcmdr)> by(DartPoints[,6:8], DartPoints$Name, rcorr.adjust)DartPoints$Name: Darl Length Width ThickLength 1.00 0.59 0.64Width 0.59 1.00 0.49Thick 0.64 0.49 1.00

n= 27

P Length Width Thick Length 0.0013 0.0003Width 0.0013 0.0095Thick 0.0003 0.0095

Adjusted p-values (Holm's method) Length Width Thick Length 0.0026 0.0009Width 0.0026 0.0095Thick 0.0009 0.0095

------------------------------------------------------------ DartPoints$Name: Pedernales Length Width ThickLength 1.00 0.40 0.52Width 0.40 1.00 0.15Thick 0.52 0.15 1.00

n= 28

P Length Width Thick Length 0.0365 0.0042Width 0.0365 0.4611Thick 0.0042 0.4611

Adjusted p-values (Holm's method) Length Width Thick Length 0.0731 0.0126Width 0.0731 0.4611Thick 0.0126 0.4611

3D Scatterplots

• Three interval/ratio variables and a possible grouping variable

• Often difficult to interpret• Experiment with rotation and view

angle• Consider dropping pins to the floor• scatter3d (car)• scatterplot3d (scatterplot3d)

with(DartPoints, scatterplot3d(Length, Width, Thick))

with(DartPoints, scatterplot3d(Length, Width, Thick, type="h"))

Bubble Plot

• Bubble plots are scatterplots in which the size of the symbol reflects a third dimension

• with(DartPoints, symbols(Length, Width, circles=Thick, inches=1/6, fg="blue", bg="blue"))

• ? symbols for more details on the variety of plots possible

> with(DartPoints, symbols(Length, Width, circles=Thick, inches=1/6, fg="blue", bg="blue"))> symbols(c(26, 26), c(35, 33), circles=c(4, 12), inches=1/6, fg="blue", bg="blue", add=TRUE)> text(c(28, 28), c(35, 33), c("Thickness = 4 mm", "Thickness = 12 mm"), pos=4)

1. Create the plot2. Add circles of min and max size to upper left3. Add text labels

Publication Quality

• In Windows, you can generally save a graph in several formats or place it in the clipboard

• For control over resolution and size, plot to a device

• Use ?Devices to get the ones available

E.g. Postscript

• postscript(file="graph.ps")• plot(rnorm(25), rnorm(25))• dev.off()

E.g. tif

• tiff(file="graph.tif", width=1500, height=1500, res=300, compression="lzw")

• plot(rnorm(25), rnorm(25), las=1)• dev.off()

Cairo

• Package Cairo gives you access to additional graphic formats including svg as well as some options that are not available in the standard graphics devices.