ML Wiki

Bar Chart

Bar Chart, or Bar Plot or Bar Graph

This is a Plot that can be useful for Exploratory Data Analysis
It's a graphical representation of Frequency Tables
- It shows the values of your data set with bars
- height of the bar is proportional to the value it represents
- so the variables you plot must be Quantitative Variables

In R

To create a bar chart in R

use barplot command

r = dnorm(seq(from=-3, to=3, length=15), mean=0, sd=1)
barplot(r, col="red")

Multivariate Analysis

Bar Charts can also be used for comparing values of two and more variables

typically, they are graphical representation of Contingency Tables

There are the following types of bar charts:

Side-by-side bar chart
- bars are put near each other
Stacked (Segmented) bar chart
- shows more information than other types - the total size, the proportion, etc
Proportional stacked bar chart
- standardized version of the stacked bar chart
- makes it easier to see the Joint Distribution of variables

In R

library(openintro)
data(email)

# stacked
t = table(email$spam, email$number)
pal = c('yellow2', 'skyblue2')
barplot(t, col=pal, beside=F)

# proportional
t.prop = rbind(t[1,] / colSums(t),
               t[2,] / colSums(t))
pal = c('yellow2', 'skyblue2')
barplot(t.prop, col=pal, beside=F)

# side-by-side
barplot(t, col=pal, beside=T)