In purpleleaf95/ggFatFonts: A ggPlot2 plug-in for FatFonts graphics

library(learnr)
knitr::opts_chunk$set(echo = FALSE)

Overview

What is this package about?

This package helps you build FatFonts graphics based on ggplot2.

FatFonts is a visualization technique which can show the symbolic as well as the quantitative aspects of numbers simultaneously, in a way that both overview and detailed views are supported. The order of each digit (i.e. in the unit¡¯s/ten¡¯s/hundred¡¯s place within a number) is revealed by the size of the whole area where the value of this digit can be represented, while the ink that is used to draw this digit in the area is determined by its exact value. There are three FatFonts types available: "Cubica", "Miguta" and "Rotunda".

{width="60%"}

__Cubica__

{width="60%"}

__Miguta__

{width="60%"}

__Rotunda__

The numbers in each plot represent 19, 28, 37, 46, 55, 64, 73, 82, 91 successively.

There are currently four quick plot functions provided: ggFFConfMatr, ggFFHeatMap, ggFFCorrMatr and ggFFClusScatPlot. They help to draw FatFonts Confusion Matrix, Heatmap, Correlation Matrix and Cluster Scatter Plot respectively. If you want to build some other plots using FatFonts, you can use addGGFFLayer function.

This package only works on Windows OS for now.

Some quick examples

Confusion Matrix

library(ggFatFonts)
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`)

{width="60%"}

Heatmap

library(ggFatFonts)
library(ggplot2)
cols <- topo.colors(9)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, changeFontColDigit = cols)

{width="60%"}

Correlation Matrix

library(ggFatFonts)
library(ggplot2)
changeFontColValueFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_color_gradient2()
  return(newPlot)
}
ggFFCorrMatr(mtcars[c(1, 3:7)], changeFontColValue = changeFontColValueFunc, adjustPos = 1, fontSize = 1)

{width="60%"}

Cluster Scatter Plot

library(ggFatFonts)
library(ggplot2)
changeBackColFunc <- function(oldPlot){
 newPlot <- oldPlot + scale_fill_gradient(low = 'white', high = 'purple')
 return(newPlot)
}
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', changeBackCol = changeBackColFunc, fontSize = 1.6)

{width="60%"}

Install ggFatFonts

Install FatFonts on your computer

There are currently three FatFonts available: "Cubica", "Miguta" and "Rotunda". Download the font files Cubica.otf, Miguta.otf and Rotunda.otf from the fonts subdirectory and then install them. If you find some difficulty in installing FatFonts, you can follow instructions on https://www.dafont.com/faq.php.

Download RStudio

Go to RStudio official site and download the latest version of RStudio. Install RStudio on your own computer by following the instructions on this website.

Download and install the package

You can directly install the package in RStudio by running the following code:

install.packages("devtools")
devtools::install_github("purpleleaf95/ggFatFonts")

Alternatively, you can download the source packages from GitHub and install it locally.

Include ggFatFonts in your session

Simply run library(ggFatFonts), and ggFatFonts will then be included in your session.

Confusion matrix

Build a FatFonts confusion matrix

The function provided in this package to produce FatFonts confusion matrix plot is ggFFConfMatr. There are two parameters that must be passed in the function: actuClass is a vector containing the actual values; predClass is a vector containing the predicted values.

Two test datasets, "confMatrTestData" and "confMatrTestData2", are built in this package to get you started. Try to build a default FatFonts confusion matrix plot using the following code:

library(ggFatFonts)
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`)

Change the position and size of FatFonts digits

Sometimes, the position of fonts may become weird using the default settings. When this happens, you can slightly adjust the position of FatFonts digits using the adjustPos parameter. The second plot looks better when this parameter has been changed from 3 (default) to 12.

library(ggFatFonts)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted, adjustPos = 12)

Another important parameter to be considered is fontSize. Automatic change of font size has been applied in this package to avoid overlapping as default (i.e. fontSize = -1). However, sometimes you may still want to manually adjust the font size. One thing to remember is that fontSize refers to the font size of the smallest digit, so do not set it to a very large number.

library(ggFatFonts)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted, adjustPos = 12, fontSize = 5)

Change font type

The default font type used in this package is ??Cubica?? (i.e. font = "Cubica"). If you want to use other FatFonts types (i.e. Miguta" or "Rotunda"), specify the font parameter. One thing to remember is that you may need to change the value of adjustPos parameter to get a better view after changing font type.

library(ggFatFonts)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted, adjustPos = 13, font = 'Rotunda', fontSize = 5)

Change shrinkage rate

Since FatFonts can only be applied to integers with limited places (i.e. usually 2 or 3 places), shrink parameter is provided to narrow the range of values within the dataset. The default value shrink = -1 refers to automatical change of shrink parameter, which will lead you to 2-level FatFonts numbers. This shrinkage rate can be found on the right bottom corner of the plot. You can change this parameter to show more or less digits.

library(ggFatFonts)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted, adjustPos = 4, shrink = 1000)

Change the aspect ratio of each grid

This function will show square grids by default (i.e. isSquare == TRUE). However, if you want to use rectangle grids, you can set isSquare == FALSE. Notice that rectangle grid is not recommended as the position of FatFonts digits may become weird due to unequal witdth and length.

library(ggFatFonts)
ggFFConfMatr(confMatrTestData2$TRUE., confMatrTestData2$predicted, adjustPos = 16, isSquare = FALSE)

Change colors

Colors of numbers, borders, backgrounds, etc. can be changed using different parameters. Details can be found in the ??Help?? panel. Instead, you can run help(ggFFConfMatr).

library(ggFatFonts)
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`, fontColor = 'blue')

You can also change the background color according to value using changeBackCol parameter. A function should be passed into this parameter, and you need to install ggplot2 package in advance.

library(ggFatFonts)
library(ggplot2)
changeBackColFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_fill_gradient(low = 'white', high = 'purple')
  return(newPlot)
}
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`, changeBackCol = changeBackColFunc)

Additionally, you can change the color of numbers according to their values using changeFontColValue parameter. A function should be passed into this parameter, and you need to install ggplot2 package in advance.

library(ggFatFonts)
library(ggplot2)
changeFontColValueFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_color_continuous(low = 'dark grey', high = 'purple')
 return(newPlot)
}
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`, changeFontColValue = changeFontColValueFunc)

Another way to change the color of numbers is to change it according to the value of digits. You can do this using changeFontColDigit parameter. A vector containing 9 colors should be passed into this parameter, and you can use color combinations provided by the system as well.

Remember that changeFontColValue and changeFontColDigit cannot be used at the same time.

library(ggFatFonts)
library(ggplot2)
cols <- topo.colors(9)
ggFFConfMatr(confMatrTestData$`actual value`, confMatrTestData$`predicted value`, changeFontColDigit = cols)

Heatmap

Build a FatFonts heatmap

The function provided in this package to produce FatFonts heatmap plot is ggFFHeatMap. There are three parameters that must be passed in the function: x is a vector containing the x values of each data point; y is a vector containing the y values of each data point; z is a vector containing the values to be displayed in FatFonts. x, y and z should be of the same length.

A test dataset, ??txhousing??, is built in this package to get you started. Try to build a default FatFonts heatmap plot using the following code:

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:3000], y = txhousing$city[1:3000], z = txhousing$sales[1:3000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales')

Customize your plot

Refer to Confusion matrix in this tutorial or the Help panel to customize your plot. Some examples are provided as follow.

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2)

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, font = 'Miguta')

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', shrink = 0.01, adjustPos = 1)

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, tileFill = 'black', fontColor = 'white')

library(ggFatFonts)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city',zAxisTitle = 'sales', fontSize = 2, isSquare = FALSE)

library(ggFatFonts)
library(ggplot2)
changeBackColFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_fill_gradient(low = 'white', high = 'purple')
  return(newPlot)
}
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, changeBackCol = changeBackColFunc)

library(ggFatFonts)
library(ggplot2)
changeFontColValueFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_color_continuous(low = 'dark grey', high = 'purple')
 return(newPlot)
}
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, changeFontColValue = changeFontColValueFunc)

library(ggFatFonts)
library(ggplot2)
cols <- topo.colors(9)
ggFFHeatMap(x = txhousing$year[1:2000], y = txhousing$city[1:2000], z = txhousing$sales[1:2000], xAxisTitle = 'year', yAxisTitle = 'city', zAxisTitle = 'sales', fontSize = 2, changeFontColDigit = cols)

Correlation matrix

Build a FatFonts correlation matrix

The function provided in this package to produce FatFonts correlation matrix plot is ggFFCorrMatr. There is one parameters that must be passed in the function - data which contains the variables displayed in columns.

The built-in dataset ??mtcars?? is introduced to get you started. Try to build a default FatFonts correlation matrix plot using the following code:

library(ggFatFonts)
ggFFCorrMatr(mtcars[,c(1,3:7)])

Change the colors of FatFonts digits

To distinguish positive and negative numbers, two rather than one font color parameters are used in this function ?C fontColorPos and fontColorNeg. You can change the value of these parameters to modify font colors.

library(ggFatFonts)
ggFFCorrMatr(mtcars[c(1, 3:7)], fontColorPos = 'blue', fontColorNeg = 'green', fontSize = 1)

Likewise, you can pass a function into the changeFontColValue parameter or a vector into thechangeFontColDigit parameter to change font colors according to the value of numbers or digits respectively. Since positive and negative numbers need to be distinguished, scale_color_gradient2() is recommended here.

library(ggFatFonts)
library(ggplot2)
changeFontColValueFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_color_gradient2()
 return(newPlot)
}
ggFFCorrMatr(mtcars[c(1, 3:7)], changeFontColValue = changeFontColValueFunc, adjustPos = 1, fontSize = 1)

library(ggFatFonts)
library(ggplot2)
cols <- topo.colors(9)
ggFFCorrMatr(mtcars[c(1, 3:7)], changeFontColDigit = cols, adjustPos = 1, fontSize = 1)

Customize your plot

Refer to Confusion matrix in this tutorial or the Help panel to customize your plot. Some examples are provided as follow.

library(ggFatFonts)
ggFFCorrMatr(mtcars[c(1, 3:7)], font = 'Miguta')

library(ggFatFonts)
ggFFCorrMatr(mtcars[c(1, 3:7)], shrink = 10, fontSize = 3)

library(ggFatFonts)
ggFFCorrMatr(mtcars[c(1, 3:7)], tileFill = 'light grey', fontSize = 1)

library(ggFatFonts)
ggFFCorrMatr(mtcars[c(1, 3:7)], isSquare = FALSE, fontSize = 1)

library(ggFatFonts)
library(ggplot2)
changeBackColFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_fill_gradient2()
  return(newPlot)
}
ggFFCorrMatr(mtcars[c(1, 3:7)], changeBackCol = changeBackColFunc, fontColorPos = 'yellow', fontColorNeg = 'yellow', fontSize = 1)

Cluster scatter plot

Build a FatFonts cluster scatter plot

The function provided in this package to produce FatFonts cluster scatter plot is ggFFClusScatPlot. There are two parameters that must be passed in the function: x is a vector containing the x values of each data point; y is a vector containing the y values of each data point. x and y should be of the same length.

A test dataset, ??mpg??, is built in this package to get you started. Try to build a default FatFonts cluster scatter plot using the following code:

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy')

Show points

If you want to show normal scatter plots on the same plot, simply specify showPoints = TRUE.

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', showPoints = TRUE)

Change bin lengths

You can change the bin length of both x and y axis by specifying the xBinLength and yBinLength parameters respectively. It is recommended to set x = y as each grid is square.

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', xBinLength = 5, yBinLength = 10, adjustPos = 1.5)

Customize your plot

Refer to Confusion matrix in this tutorial or the Help panel to customize your plot. Some examples are provided as follow.

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', font = 'Rotunda', adjustPos = 1.1)

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', tileFill = 'black', fontColor = 'white')

library(ggFatFonts)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', isSquare = FALSE, adjustPos = 1.5)

library(ggFatFonts)
library(ggplot2)
changeBackColFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_fill_gradient(low = 'white', high = 'purple')
  return(newPlot)
}
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', changeBackCol = changeBackColFunc)

library(ggFatFonts)
library(ggplot2)
changeFontColValueFunc <- function(oldPlot){
  newPlot <- oldPlot + scale_color_continuous(low = 'dark grey', high = 'purple')
 return(newPlot)
}
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', changeFontColValue = changeFontColValueFunc)

library(ggFatFonts)
library(ggplot2)
cols <- topo.colors(9)
ggFFClusScatPlot(mpg$cty, mpg$hwy, xAxisTitle = 'cty', yAxisTitle = 'hwy', changeFontColDigit = cols)

Advanced application

If you are familiar with ggplot2, you can use the addGGFFLayer function provided in this package together with the functions provided in ggplot2 package to build your own FatFonts plots. The most important thing to do in this case is to change the data structure so that it can be read by both ggplot2 and addGGFFLayer. An example is provided as follow:

# Library
library(ggplot2)
library(ggFatFonts)

# Creat test dataset
test <- data.frame(var1 = c('a', 'a', 'b', 'b'),
var2 = c('a', 'c', 'a', 'd'),
fill = c(12, 53, 87, 6))

# Create ggplot without FatFonts layer
myPlot <- ggplot(data = test, mapping = aes(x = var1, y = var2, fill = fill)) +
geom_tile() +
coord_fixed(ratio = 1, expand = FALSE)

# Add FatFonts layer
myPlot <- addGGFFLayer(oldPlot = myPlot,
data = test[, 3],
xBinLength = 1,
yBinLength = 1,
font = 'Cubica',
fontSize = 8,
fontColor = '#FFFFFF',
adjustPos = 9)
print(myPlot)

purpleleaf95/ggFatFonts documentation built on May 15, 2019, 1:18 p.m.