Introduction to data Visualization in R

**Nilanjan Chatterjee** February, 2020

Topics

• What is data visualization?
• Why is data visualization important?
• How to do data visualization?
• Possible options and pitfalls

What is data visualization?

Technique to communicate insights from data through visual representation.
Allow easy understanding of large dataset.
Provides basic knowledge about variables.
Most efficient way to identify, locate, manipulate, format, and present data.

Why data visualization is important?

• Ever increasing amount of data.
• Humanly impossible to see distinct patterns.
• Improved insight.
• Faster Decision making.

How to do data visualization?

• Plot in Base R
• ggplot2 package and associates

data(mtcars)
plot(mpg~wt, mtcars, pch=19, col="blue")

plot vs ggplot

plot vs ggplot

Pros	Cons
In-built	Additional package
Easy to learn	Steep learning curve
Indepenedent of data-structures	Works only with data-frame
Easy for simple plots	Verbose for complex plots
Low level of abstraction	High abstraction level
Visually less appealing	Visually more appealing

ggplot

ggplot

Based on Grammer of graphics (Wilkinson, 2005).
Consists of several building blocks like a sentence.

• data
• aesthetic mapping
• geometric object
• scales
• coordination system
• position adjustmnets
• faceting

ggplot

#install.packages("ggplot2", dependencies = T)
library(ggplot2)
ggplot(mtcars, aes(x= wt, y= mpg))+ 
  geom_point(colour="blue", size=3) 

How to plot in ggplot

ggplot(mtcars) #data

How to plot in ggplot

ggplot(mtcars, aes(x= wt, y= mpg)) #data+aesthetic map

How to plot in ggplot

ggplot(mtcars, aes(x= wt, y= mpg))+ #data+aesthetic map
  geom_point() #geometric obj

How to plot in ggplot

ggplot(mtcars, aes(x= wt, y= mpg))+ #data+aesthetic map
  geom_point(colour="blue", size=3) #geometric obj

How to plot in ggplot

ggplot(mtcars, aes(x= wt, y= mpg))+ #data+aesthetic map
  geom_point(colour="blue", size=3)+ #geometric obj
  ggtitle("Scatterplot") #Plot title

Different sections of ggplot

• DATA only data-frame is allowed
• AES takes into account the aesthetics
• GEOM stands for the different geometrices
  • geom_point for point plot
  • geom_bar for barplot
  • geom_line for line plot
  • geom_histogram for histogram
  • geom_boxplot for boxplot
    and so on

Some more examples

ggplot(mtcars, aes(x=mpg))+
  geom_bar()

Some more examples

ggplot(mtcars, aes(x=cyl, y=mpg, fill= cyl))+ 
  geom_bar(stat="identity")

Some more examples

ggplot(mtcars, aes(x=cyl, y=mpg))+ 
  geom_point(stat="identity", size=4)

Export graphs from R/Rstudio

You can export any plots using the plot window from R/RStudio.
To save files in high-resolution these commands are helpful

sct <-ggplot(mtcars, aes(x= wt, y= mpg))+ 
  geom_point(colour="blue", size=3)+   ggtitle("Scatterplot")

ggsave(sct, "Scatterplot_with_R.jpeg", dpi=100, device = "jpeg")

Exercise

• Use your own data and make a basic plot (scatterplot, barplot, histogram) in ggplot
• change the color of the plot
• What is the difference if you put colour or shape in data part rather than geometric object part?

Thanks

• For further queries nilanjan@wii.gov.in
• Slides: https://nilanjanchatterjee.github.io/projects/