1. Color, legends, and labeling
2. Basic principles of part-to-whole
3. Examples of basic methods to visualize proportions
4. Additional methods to visualize proportions
5. Interactive visualizations

• We can use the scale_? functions to change things like:
  • Legend labeling
  • Color palettes
  • Scaling properties of the variable
• These functions generally have three parts to their name:
  • ‘scale’
  • The visual attribute being modified (x, y, color, fill, shape, size, etc.)
  • The type of modification (continuous, discrete, manual, gradient, etc.)

• For example:
  • scale_x_continuous – modifies the visual element that encodes continous values by positioning along the x-axis
  • scale_color_discrete – modifies the visual element that encodes discrete numeric values by assigning them a border color
  • scale_fill_manual – modifies the visual element that encodes categorical values by assigning them a fill color
  • scale_fill_brewer – modifies fill color using the RColorBrewer library

• For information about changing legends, see: http://www.cookbook-r.com/Graphs/Legends_(ggplot2)

• For general ggplot2 color information, see: http://www.cookbook-r.com/Graphs/Colors_(ggplot2)

• For more information about RColorBrewer, see: http://earlglynn.github.io/RNotes/package/RColorBrewer/index.html

library(ggplot2)

ggplot(data=PlantGrowth, aes(x=group, y=weight, fill=group)) + 
  geom_boxplot() +  
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)

ggplot(data=PlantGrowth, aes(x=group, y=weight, fill=group)) + 
  geom_boxplot() +  
  scale_fill_manual(values=c("#999999", "#E69F00", "#56B4E9"), 
                       name="Experimental\nCondition",
                       breaks=c("ctrl", "trt1", "trt2"),
                       labels=c("Control", "Treatment 1", "Treatment 2")) +
   xlab("") + ylab("Weight (g)") +
   theme(text=element_text(size=18,family="Times"))

myMlotObj + guides(fill = guide_legend(reverse=TRUE))

myMlotObj + scale_fill_discrete(guide = guide_legend(reverse=TRUE))

myMlotObj + scale_fill_discrete(breaks = rev(levels(PlantGrowth$group)))

library(ggplot2)

df2 = read.table('http://eecs.ucf.edu/~wiegand/idc6700/datasets/color-cookbook-eg.txt', header=T)
ggplot(df2, aes(x=cond1, y=yval)) + 
    geom_line(aes(colour=cond2, group=cond2)) +
    geom_point(aes(colour=cond2), size=3)  +
    theme(text=element_text(size=18,family="Times"))

## List of 1
##  $ text:List of 11
##   ..$ family       : chr "Times"
##   ..$ face         : NULL
##   ..$ colour       : NULL
##   ..$ size         : num 18
##   ..$ hjust        : NULL
##   ..$ vjust        : NULL
##   ..$ angle        : NULL
##   ..$ lineheight   : NULL
##   ..$ margin       : NULL
##   ..$ debug        : NULL
##   ..$ inherit.blank: logi FALSE
##   ..- attr(*, "class")= chr [1:2] "element_text" "element"
##  - attr(*, "class")= chr [1:2] "theme" "gg"
##  - attr(*, "complete")= logi FALSE
##  - attr(*, "validate")= logi TRUE

library(ggplot2)

df2 = read.table('http://eecs.ucf.edu/~wiegand/idc6700/datasets/color-cookbook-eg.txt', header=T)
cbPalette <- c("#999999", "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#D55E00", "#CC79A7")
ggplot(df2, aes(x=cond1, y=yval)) + 
    geom_line(aes(color=cond2, group=cond2), size=1.15) +
    geom_point(aes(color=cond2), size=4)  +
    scale_color_manual(values=cbPalette,
                       breaks=c("J","I","K","L"),
                       name="Condition 2") +
    xlab("Condition 1") + ylab("y-Value")
    theme(text=element_text(size=18,family="Times"))

## List of 1
##  $ text:List of 11
##   ..$ family       : chr "Times"
##   ..$ face         : NULL
##   ..$ colour       : NULL
##   ..$ size         : num 18
##   ..$ hjust        : NULL
##   ..$ vjust        : NULL
##   ..$ angle        : NULL
##   ..$ lineheight   : NULL
##   ..$ margin       : NULL
##   ..$ debug        : NULL
##   ..$ inherit.blank: logi FALSE
##   ..- attr(*, "class")= chr [1:2] "element_text" "element"
##  - attr(*, "class")= chr [1:2] "theme" "gg"
##  - attr(*, "complete")= logi FALSE
##  - attr(*, "validate")= logi TRUE

library(ggplot2)
library(RColorBrewer)

df2 = read.table('http://eecs.ucf.edu/~wiegand/idc6700/datasets/color-cookbook-eg.txt', header=T)
ggplot(df2, aes(x=cond1, y=yval)) + 
    geom_line(aes(color=cond2, group=cond2), size=1.15) +
    geom_point(aes(color=cond2), size=4)  +
    scale_color_brewer(palette="Set1",
                       breaks=c("J","I","K","L"),
                       name="Condition 2") +
    xlab("Condition 1") + ylab("y-Value")
    theme(text=element_text(size=18,family="Times"))

library(ggplot2)

myScatterData = data.frame(Meeples=runif(20),
                           Furples=runif(20),
                           Troddles=runif(20))

ggplot(myScatterData, aes(x=Meeples, y=Furples, fill=Troddles)) + 
    geom_point(size=4, shape=21) +
    scale_fill_gradient(low="yellow", high="red") +
    theme(text=element_text(size=18,family="Times"))

library(ggplot2)

myScatterData = data.frame(Meeples=runif(20),
                           Furples=runif(20),
                           Troddles=runif(20))

ggplot(myScatterData, aes(x=Meeples, y=Furples, size=Troddles)) + 
    geom_point() +
    scale_size_continuous(range=c(1,20)) +
    theme(text=element_text(size=18,family="Times"))

There are a variety of typical part-to-whole patterns:

• All values are roughly the same
• Differences from one value to the next change by roughly the same amount
• Differences from one value to the next decrease
• Differences from one value to the next increase
• Differences from one value to the next change vary significantly
• Differences from one value to the next start small, become larger, then small again
• One value is exceptionally larger (or smaller) than the others

• Pie charts and donut charts
• Bar graphs
• Dot plots
• Pareto Charts

• Consider grouping categorical items in different ways
• Use percentile scales for Pareto charts
• Re-express variables if there are scaling issues (e.g., use log scale)
• Use line graphs to view ranking changes over time

• Consider the mpg dataset
• In particuar: proportions for categorical data for car class and drive type
• Tour methods we’ve used before to show this same data in different ways

1. Bar Chart, Stacked Bar Chart
2. Pie Chart, Donut Chart
3. Multivariate Fixed-Area Stacked Bar Chart Source
4. Multivariate, Trellised Bar Chart
5. Multivariate, Trellised Pie Chart

library(ggplot2)

ggplot(mpg, aes(x=class)) + 
  geom_bar() +  
  xlab("Vehicle Class") +
  ylab("Count") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)
library(RColorBrewer)

numLevels = length(levels(factor(mpg$class)))
classPalette = colorRampPalette(brewer.pal(5,"Blues"))(numLevels)

ggplot(mpg, aes(x=factor(1),fill=class)) + 
  geom_bar() +  
  scale_fill_manual(values=classPalette) +
  xlab("Vehicle Class") +
  ylab("Count") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)
library(dplyr)

mpg.ordered = summarise(group_by(mpg, class), Count=length(class))

ggplot(mpg.ordered, aes(x=reorder(class, -Count), y=Count)) + 
  geom_bar(stat="identity", fill="darkblue") +  
  xlab("Vehicle Class") +
  ylab("Count") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)
library(dplyr)

totalCount = dim(mpg)[1]
mpg.ordered = summarise(group_by(mpg, class), Count=length(class))
mpg.ordered = arrange(mpg.ordered, -Count)
mpg.ordered$Percentage = 100 * mpg.ordered$Count / totalCount
mpg.ordered$Remainder = 100 * cumsum(mpg.ordered$Count) / totalCount

ggplot(mpg.ordered, aes(x=reorder(class, -Count), y=Percentage)) + 
  geom_bar(stat="identity", fill="darkgray") +  
  geom_path(aes(y=Remainder, group=1), size=1.25, color="firebrick") +
  geom_point(aes(y=Remainder), size=4) +
  xlab("Vehicle Class") +
  ylab("Percentage") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)
library(RColorBrewer)

numLevels = length(levels(factor(mpg$class)))
classPalette = colorRampPalette(brewer.pal(5,"Blues"))(numLevels)

ggplot(mpg, aes(x="",fill=class)) + 
  geom_bar() +  coord_polar(theta="y") +
  scale_fill_manual(values=classPalette) +
  xlab("") +
  ylab("") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"),
        axis.ticks.y = element_blank(),
        axis.text.y = element_blank())

library(ggplot2)
library(RColorBrewer)
library(reshape2)

# Create the color ramp
numLevels = length(levels(factor(mpg$class)))
classPalette = colorRampPalette(brewer.pal(5,"Blues"))(numLevels)

# Create the totals for the rectangle y positions
classCounts = mutate(melt(table(mpg$class), value.name="Count"),
                     ymax=cumsum(Count))
classCounts$ymin = c(0, head(classCounts$ymax, length(classCounts$ymax)-1))

# Plot this thing
ggplot(classCounts, aes(fill=Var1, ymin=ymin, ymax=ymax, xmin=3, xmax=4)) + 
  geom_rect() +  
  coord_polar(theta="y") +
  scale_fill_manual(values=classPalette, name="Car  Class") +
  xlim(c(0,4)) +
  xlab("") +
  ylab("") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"),
        axis.ticks.y = element_blank(),
        axis.text.y = element_blank())

library(ggplot2)
library(RColorBrewer)

numLevels = length(levels(factor(mpg$class)))
classPalette = colorRampPalette(brewer.pal(5,"Blues"))(numLevels)

ggplot(mpg, aes(x=drv,fill=class)) + 
  geom_bar(position="fill") +  
  scale_fill_manual(values=classPalette) +
  xlab("Drive") +
  ylab("Ratio") +
  ggtitle("Vehicles of Different Classes & Drives, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)

labels = c("4"="Four Wheel Drive", "f"="Front Wheel Drive", "r"="Rear Wheel Drive")

ggplot(mpg, aes(x=class)) + 
  geom_bar() +  coord_flip() +
  facet_grid(facets=drv ~ ., labeller=labeller(drv=labels)) +
  xlab("Vehicle Class") +
  ylab("Count") +
  ggtitle("Vehicles of Different Classes & Drives, 1999-2008") +
  theme(text=element_text(size=18,family="Times"))

library(ggplot2)
library(RColorBrewer)

numLevels = length(levels(factor(mpg$class)))
classPalette = colorRampPalette(brewer.pal(5,"Blues"))(numLevels)

labels = c("4"="Four Wheel Drive", "f"="Front Wheel Drive", "r"="Rear Wheel Drive")

ggplot(mpg, aes(x=factor(1),fill=class)) + 
  geom_bar(position="fill") +  coord_polar(theta="y") +
  facet_grid(facets=. ~ drv, labeller=labeller(drv=labels)) + 
  scale_fill_manual(values=classPalette) +
  xlab("Drive") +
  ylab("") +
  ggtitle("Vehicles of Different Classes, 1999-2008") +
  theme(text=element_text(size=18,family="Times"),
        axis.ticks.y = element_blank(),
        axis.text.y = element_blank())

library(ggplot2)
library(RColorBrewer)
library(gcookbook) # for uspopage

ggplot(uspopage, aes(x=Year, y=Thousands, fill=AgeGroup)) +
  geom_area(color="black", size=0.2, alpha=0.4, position="fill") +
  scale_fill_brewer(palette="Blues", breaks=rev(levels(uspopage$AgeGroup))) +
  theme(text=element_text(size=18,family="Times")) +
  ylab("Proportion of People in Age Group") +
  ggtitle("Proportion of People of Different Ages in US, 1900-2000")

library(ggplot2)
library(dplyr)
library(reshape2)
library(RColorBrewer)

# Variables to setup the filter
semester="fall"
year=2017
courseID="DIG5876"
instructorName="Rudolf Wiegand"

# Get the data
offset =c("spring"=0, "summer"=10, "fall"=20)
semesterID = 30*(year-1996)+960 + offset[semester]
dataFile = paste('Data/ucfspi',semesterID,'.data', sep='')
spi <- read.delim(dataFile, sep='#', header=F)
rpwSPI = filter(spi, InstructorName == instructorName & CourseID == courseID)

# Column/Variable Names
colnames(spi) <- c("College", "Department", "InstructorEmail", "InstructorName", 
                   "CourseName", "CourseID", "CourseSuffix", "SemesterID", 
                   "n",  "Organization", "Expectations", "Communicating",
                   "Respect", "Interest", "LearningEnv", "PerformanceFeedback",
                   "AchieveObjectives", "OverallEffectiveness", "NotUsed")

# Put the data in long form
rpwSPI.melt = melt(rpwSPI, id=1:9)

# We only care about the questions and make sure the responses
# are treated like an ordered factor (ordinal data), not numeric
rpwSPI.corrected = mutate(filter(rpwSPI.melt, variable != "NotUsed"), 
                          Responses = factor(value, levels=factor(5:1),
                                             ordered=T))

# Tally up all 1's, 2's, 3's, 4's, and 5's for each question
rpwSPI.counts = summarise(group_by(rpwSPI.corrected, Responses, variable),
                          Count=length(Responses))

# Get overall counts in order to obtain percentages
rpwSPI.totals = summarise(group_by(rpwSPI.corrected, variable),
                          Count=length(variable))
overallNN = max(rpwSPI.totals$Count)

# Setup some question labels
questions = c("Effectiveness organizing the course",
              "Effectiveness explaining course requirements,\n grading criteria, and expectations",
              "Effectiveness communicating ideas\n and/or information",
              "Effectiveness showing respect and\n concern for student",
              "Effectiveness stimulating interest\n in the course",
              "Effectiveness creating an environment\n that helps students learn",
              "Effectiveness giving useful feedback\n on course performance",
              "Effectiveness helping students\n achieve course objectives",
              "Overall effectiveness of the instructor")

# Setup legend texts and palette
responseText = c("Strongly Agree",
                 "Agree",
                 "No Opinion",
                 "Disagree",
                 "Strongly Disagree")

likertPalette = c("#336633",  # Strongly Agree, 5
                  "#99CC99",  # Agree, 4
                  "#999999",  # No Opinion, 3
                  "#FF6666",  # Disagree, 2
                  "#FF3333")  # Strongly Disagree, 1

# The plot!
ggplot(rpwSPI.counts, aes(x=variable, y=100*Count/overallNN, fill=Responses)) + 
  geom_bar(stat="identity") +
  geom_hline(yintercept=c(0,25,50,75,100), color="white", size=1) +
  scale_fill_manual(values=likertPalette,
                    name = "",
                    breaks = 5:1,
                    labels=responseText) +
  scale_x_discrete(labels=questions) +
  coord_flip() + 
  ylab("Percentage") +
  xlab("") +
  guides(fill = guide_legend(reverse=T)) +
  theme_bw() +
  theme(text=element_text(family="Times", size=18),
        axis.text.y = element_text(size=16, hjust=1),
        legend.position = "bottom",
        panel.grid.major.x = element_blank(),
        panel.grid.minor.x = element_blank(),
        panel.border = element_blank(),
        panel.background = element_blank())

library(portfolio)
posts <- read.csv("http://datasets.flowingdata.com/post-data.txt")
map.market(id=posts$id, 
           area=posts$views, 
           group=posts$category, 
           color=posts$comments, 
           main="FlowingData Map")

library(tm)
library(wordcloud)

rpwDissertationURL = "http://cs.ucf.edu/~wiegand/ids6938/datasets/rpw-diss.txt"
dissertation = Corpus(URISource(rpwDissertationURL))

# Take out all white space and change everything to lower case
dissertation = tm_map(dissertation,stripWhitespace)  
dissertation = tm_map(dissertation, tolower)

# Remove unnecessary and redundant words with slight suffix variations
dissertation = tm_map(dissertation, removeWords, stopwords("english"))
dissertation = tm_map(dissertation, stemDocument)

# Make sure it is really a text document
dissertation = tm_map(dissertation, PlainTextDocument) 

wordcloud(dissertation,
          scale=c(5,0.5), 
          max.words=150, 
          random.order=FALSE, 
          rot.per=0.35)

library(ggplot2)
library(dplyr)
library(mapproj)

# Extract reference data
mapstates <- map_data("state")

# Get the electoral data
elec2008URL = "http://www.electoral-vote.com/evp2008/Pres/Final-2008.csv"
elec2008 = mutate(read.csv(elec2008URL),
                  region=tolower(as.character(State)))

# Merge the data for the chloropleth
plotableCountyData = merge(mapstates, elec2008)

ggplot(plotableCountyData, aes(long,lat,group=group)) + 
  geom_polygon(aes(fill=Obama.Pct)) + 
  scale_fill_gradient(low="white",high="darkblue", limits=c(0,100)) +
  coord_map(project="globular") +
  geom_path(data = mapstates, colour = "white", size = .75, alpha = .1) +
  ggtitle("Who Voted for Obama in 2008?")

library(ggplot2)
library(dplyr)
library(mapproj)

# Extract reference data
mapstates <- map_data("state")

# Get the electoral data
elec2008URL = "http://www.electoral-vote.com/evp2008/Pres/Final-2008.csv"
elec2008 = mutate(read.csv(elec2008URL),
                  region=tolower(as.character(State)))

# Merge the data for the chloropleth
plotableCountyData = merge(mapstates, elec2008)

ggplot(plotableCountyData, aes(long,lat,group=group)) + 
  geom_polygon(aes(fill=Obama.Pct)) + 
  scale_fill_gradient(low="white",high="darkblue", limits=range(plotableCountyData$Obama.Pct)) +
  coord_map(project="globular") +
  geom_path(data = mapstates, colour = "white", size = .75, alpha = .1) +
  ggtitle("Who Voted for Obama in 2008?")

• A chloropleth is a geographic map where a numeric value is encoded as a color for a region of that map to see how a variable relates to a geography
• A heat map is a tiled data visualization to help show relationships between variables
• There are many visualizations that use geometric shapes and colors over regions; most of them are not heatmaps
• We’ll learn more about heatmaps later