Visualizing data - with R

"Data visualization refers to the techniques used to communicate data or information by encoding it as visual objects (e.g., points, lines or bars) contained in graphics." Wikipédia

Data visualization is part of the Data Science Process

R for data science - Garrett Grolemund & Hadley Wickham

"R is a free software environment for statistical computing and graphics." The R Project

"R is a free software environment for statistical computing and graphics." The R Project

1+1

## [1] 2

"R is a free software environment for statistical computing and graphics." The R Project

1+1

## [1] 2

plot(1:10, 1:10)

• Ecosystem - Pipeline: import / clean / transform / analyse / calculate / modelize / visualize / report
• Free and Open Source
• Reproducibility:
  • Share the data analysis process, not just the final product
  • Validation of results by others
  • Re-run analysis when data changes
  • share, edit, remix...
• > 10K Librairies
• Very active community
• Strong graphic capabilities

Population, Life expectency and Gross Domestic Product per capita for 142 countries and 12 years.

library(gapminder)
head(gapminder)

The Gapminder Library

data=subset(gapminder, year==2007)
plot(data$lifeExp ~ data$gdpPercap)

Take a few minutes: what would you do to communicate this result? ...

Take a few minutes: what would you do to communicate this result? ...

plot(data$lifeExp ~ data$gdpPercap,
     xlab="Gdp per capita", ylab="Life Expectancy",
     main="Features of countries in 2007")

plot(data$lifeExp ~ data$gdpPercap,
     xlab="Gdp per capita", ylab="Life Expectancy",
     main="Features of countries in 2007",
     pch=20, cex=3)

plot(data$lifeExp ~ data$gdpPercap,
     xlab="Gdp per capita", ylab="Life Expectancy",
     main="Features of countries in 2007",
     pch=20, cex=3, col="blue")

# attribute a color to each continent:
my_colors=pal[as.numeric(data$continent)]

# use this vector as color for the plot
plot(lifeExp ~ gdpPercap, data=data, pch=20, cex=3, col=my_colors)

#plot
my_colors=pal[as.numeric(data$continent)]
plot(lifeExp ~ gdpPercap, data=data, pch=20, cex=3, col=my_colors)

#add legend
legend("bottomright", legend=levels(data$continent), col=pal, pch=20, bty="n", pt.cex=3, horiz = F)

# Map the color:
library(RColorBrewer)
pal <- brewer.pal(5, "Set1")
my_colors=pal[as.numeric(data$continent)]

# Make the plot 
par(mar=c(3,3,2,2)) # Margin
plot(data$lifeExp ~ data$gdpPercap,

     # titles
     xlab="Gross Domestic Product per capita", ylab="Life Expectancy",
     main="Features of countries in 2007",

     # color
     col=my_colors,

     # shapes
     pch=20, cex=3,

     # no box:`
     bty="l")

#add legend
legend("bottomright", legend=levels(data$continent), col=pal, pch=20, bty="n", pt.cex=3, horiz = F)

"Help and Inspiration concerning R graphics"

www.r-graph-gallery.com

Propose or request a graph

library(ggplot2)
ggplot(data, 
  aes(gdpPercap, lifeExp, size = pop, color = continent, frame = year)) +
  geom_point()

• Created by Hadley Wickham in 2005
• Based on Leland Wilkinson's book: The Grammar of Graphics
• "tries to take the good parts of base and lattice graphics and none of the bad parts"
• 2 modes: qplot() and ggplot()
  
  

Jennybc - ggplot2 tutorial

• The R Graphics Cookbook by Winston Chang
• The ggplot2: Elegant Graphics for Data Analysis by Hadley Wickham
• Learn with examples with the Ggplot2 section of the R graph gallery
  
  

ggplot(gapminder, aes(gdpPercap, lifeExp, size = pop, color = continent, frame = year)) +
  geom_point() +
  xlim(0, 60000) +
  facet_wrap(~year)

ggplot(data, aes(gdpPercap, lifeExp, size = pop, color = continent, frame = year)) +
  geom_point() +
  xlim(0, 60000) +
  facet_wrap(~continent, nrow=3) + 
  theme(legend.position="none")

ggplot(gapminder, aes(x=continent, y=lifeExp, color=continent, fill=continent)) + 
  geom_boxplot(alpha=0.3)  + 
  theme(legend.position="none")

ggplot(gapminder, aes(x=continent, y=lifeExp, color=continent, fill=continent)) + 
  geom_violin(alpha=0.3)  + 
  theme(legend.position="none")

ggplot(gapminder, aes(x=continent, y=lifeExp, color=continent, fill=continent)) + 
  geom_boxplot(alpha=0.3) + 
  geom_jitter(color="grey", size=0.8) + 
  theme(legend.position="none")

library(dplyr)
gapminder  %>% 

    select(continent, year, pop) %>% 
    group_by(year, continent) %>% 
    summarize(sum_pop = sum(as.numeric(pop))) %>% 

    ggplot( aes(fill=continent, y=sum_pop, x=year)) + 
        geom_bar(stat="identity") + 
        ylab("Population per continent")

library(dplyr)
gapminder  %>% 
    filter(continent=="Asia") %>% filter(pop > 50000000) %>%
    select(country, year, pop) %>% 
    group_by(year, country) %>% 
    ggplot( aes(x=year, y=pop, color=country, fill=country)) + 
        geom_area() +
        facet_wrap(~country)+
        theme(legend.position="none")

• Zoom on a specific part
• Get information when hovering
• Make groups appear / disappear
• export directly
• move on axis
• Play with your chart
• Make your dataviz alive!

The HTML Widgets for R

• "Plotly is the modern platform for agile business intelligence and data science"
• https://plot.ly/
• and a html widget as well

library(plotly)

• Make a plot with plot_ly() or ggplotly()

# Basic ggplot2 chart
p=ggplot(data, 
  aes(gdpPercap, lifeExp, size = pop, color = continent, text=country)) +
  geom_point()

# Made interactive with plotly
library(plotly)
ggplotly(p)

If you know ggplot2, you know how to do interactive charts!

• Copy and paste in powerpoint? in an e-mail?
• Make a figure for publication with handmade modification?
• Are you sure you can provide exactly the same result as last time?
  
  

• Copy and paste in powerpoint? in an e-mail?
• Make a figure for publication with handmade modification?
• Are you sure you can provide exactly the same result as last time?
  
  

Plos Medicine

We need reproductibility, And R is the perfect tool for that.

• Turn your analysis into reports
• Fully reproducible
• Weave together narrative text and code
• Many output formats: PDF, HTML, websites...

http://rmarkdown.rstudio.com/

---
title: "Analysing the Gapminder dataset"
author: "Yan Holtz"
date: '`r as.character(format(Sys.Date(), format="%d/%m/%Y"))`'
output:
  html_document:
    toc: yes
---

---
title: "Analysing the Gapminder dataset"
author: "Yan Holtz"
date: '`r as.character(format(Sys.Date(), format="%d/%m/%Y"))`'
output:
  html_document:
    toc: yes
---


# 1- Introduction
Hi Robert, here is my reproducible analysis concerning the Gapminder dataset!

---
title: "Analysing the Gapminder dataset"
author: "Yan Holtz"
date: '`r as.character(format(Sys.Date(), format="%d/%m/%Y"))`'
output:
  html_document:
    toc: yes
---


# 1- Introduction
Hi Robert, here is my reproducible analysis concerning the Gapminder dataset!


# 2- Get data
The data are included in the gapminder library
\```{r}
library(gapminder)
head(gapminder)
\```

See Code See HTML document

See Code See HTML document

Gaston Sanchez

Alisa Singer

Alisa Singer

Alisa Singer

Marcus Volz

Marcus Volz

Alisa Singer

Nadieh Bremer

Olivier O'Brien

Yan Holtz

Victor Perrier

Yan Holtz
yan1166@hotmail.com
holtzyan.wordpress.com
Slide made with Slidify
And available on github.com/holtzy