Description
Exercise 1:
Obtain a heatmap based on the USArrests data set, and interpret the plot.
Specifically, comment on which states have higher and lower values of these statistics, comment on any
apparent relationships between these crime rates and urban population, and comment on groups of
states that are the most similar with respect to these statistics and groups of states that are very
different with respect to this statistics.
Exercise 2:
The state data set in R contains some basic information about the states in the United States in 1975.
The states are in alphabetical order in the state data set and in the USArrests data set, so it is easy
to directly combine information from them.
Combine the USArrests data with the region and abbreviation (abb) variables from the state data
set to create a data set for the treemap exercises.
Use a treemap to compare state-by-state urban population and assault rates, grouping by region within
the plot.
Comment on urban populations and assault rates by region and within region. What were the general
differences between regions at the time? What were the differences within regions? Which states had
the highest and lowest urban populations and assault rates overall in 1973, and which were highest and
lowest within regions.
Exercise 3:
Repeat Exercise 2 with murder and assault rates instead of urban population and assault rates.
Exercise 4:
Create visualizations of the following three kernel density estimations.
• Murder rate and assault rate
• Murder rate and urban population
• Assault rate and urban population
Be sure to choose bandwidths that provide reasonable refinement for the density estimation (not overly
smoothed or under-smoothed). Comment on what the estimations tell us about more common
magnitudes of the arrest rates and urban populations at the time, and what the plots tell us about
relationships between the arrest rates and between the arrest rates and urban population in 1973.
Additional Exercise for Graduate Students
Exercise 5:
Obtain and compare visualizations of murder and assault rate kernel density estimations by region using
the region variable added for the treemaps. Comment on how the arrest rates differ across regions and
how the individual regions compare with the overall density estimation for murder and assault rates
from Exercise 4.
Get the plots by group programmatically rather than doing each plot in separate replicated code.
STAT 480
