Spatial Data Science MOOC

Exercise
Explore data using data visualization
techniques
Section 1 Exercise 2
November 3, 2020
 Spatial Data Science MOOC

Explore data using data visualization techniques

Time to complete
80 minutes

Introduction
Data visualization helps you digest information by using symbols to visually represent
quantities and categories. You can quickly make comparisons and perceive relative
proportions, patterns, relationships, and trends. Data visualization is important throughout the
analysis process, from exploring your data, to interpreting your results, to communicating your
findings. There are various data visualization techniques available in ArcGIS. In this exercise,
you will use these techniques to explore your data and look for any interesting relationships
that may be useful in a predictive analysis.

Exercise scenario
Because voting is voluntary in the United States, the level of voter participation (referred to as
"voter turnout") has a significant impact on the election results and resulting public policy.
Modeling voter turnout, and understanding where low turnout is prevalent, can inform
outreach efforts to increase voter participation. In this exercise, you will use various
visualization techniques to explore relationships and patterns of voter turnout and to identify
potential variables to use in your predictive analysis.

Step 1: Open an ArcGIS Pro project

To begin, you will open the ArcGIS Pro project package that you downloaded for the first
exercise in this section.

a Start ArcGIS Pro.

b If necessary, sign in using the provided course ArcGIS account.

c In the bottom-left corner of the Start page, click Open Another Project.
Note: If you have configured ArcGIS Pro to start without a project template or with a default
project, you will not see the Start page. On the Project tab, click Open, and then click Open
Another Project.

Copyright © 2020 Esri. All rights reserved. 1

 Spatial Data Science MOOC

d Browse to the DataEngineering_and_Visualization folder that you saved on your

computer.

e Click DataEngineering_and_Visualization.aprx to select it, and then click OK.

f In the Catalog pane, expand Maps.

g Right-click Data Visualization and choose Open.

A Data Visualization map tab opens to a gray basemap with a map layer that contains the
2016 election results for each county.

Step 2: Explore data attributes and visualize distributions

The 2016 election results layer (CountyElections2016) includes voter turnout and
demographic variables geoenriched to the layer.

Copyright © 2020 Esri. All rights reserved. 2

 Spatial Data Science MOOC

a In the Contents pane, right-click CountyElections2016 and choose Attribute Table.

The CountyElections2016 attribute table appears below the map.

b In the attribute table, scroll to the right to familiarize yourself with the variables included
in the dataset.

c Scroll back to the left, if necessary, and then right-click the Voter_Turnout attribute column
and choose Statistics.

Note: Your figures may be slightly different in this step.

A chart window displaying a histogram appears below the map, and the Chart Properties
pane opens.
The histogram visually summarizes the distribution of voter turnout by measuring the
frequency at which values appear in the dataset. The x-axis represents different ranges, or
bins, of voter turnout values. The y-axis measures the number of counties with voter turnout
values falling within each range. To learn more about histograms in ArcGIS Pro, see ArcGIS
Pro Help: Histogram.

d Examine the histogram Chart Properties pane.

The Chart Properties pane includes a list of statistics summarizing the Voter_Turnout variable.
The mean county voter turnout value is about 0.59, with county values ranging from
approximately 0.16 to approximately 0.94. By default, a line representing the mean value is
overlaid on the histogram. You can add additional overlays from the Chart Properties pane.

Copyright © 2020 Esri. All rights reserved. 3

 Spatial Data Science MOOC

e In the Chart Properties pane, under Variable, check the Show Normal Distribution box.

f Under Statistics, check the Median and Std. Dev. boxes.

The overlays appear on the histogram. Using these overlays, you can see that the histogram
has a fairly symmetrical bell shape with nearly identical mean and median values. This
outcome indicates that the Voter_Turnout variable is normally distributed. In a normal
distribution, the mean, median, and mode values are equal. This outcome means that most
values fall near the average in the center of the distribution, with fewer and fewer values
appearing as you move farther from the center into the left and right tails.
You can also use the histogram to interactively select features on the map based on their
voter turnout values.

g In the chart window, drag a box around the bins with the lowest voter turnout values in
the left tail of the histogram, as shown in the following graphic.

Copyright © 2020 Esri. All rights reserved. 4

 Spatial Data Science MOOC

A gray box appears to indicate which bins will be selected.

Copyright © 2020 Esri. All rights reserved. 5

 Spatial Data Science MOOC

The counties with the lowest voter turnout are highlighted in the histogram and in the map.
You can review these records in the attribute table.

Copyright © 2020 Esri. All rights reserved. 6

 Spatial Data Science MOOC

h Click and hold the CountyElections2016 attribute table tab.

i Drag the attribute table tab to the center of the map until the docking target appears, as
shown in the following graphic.

The attribute table is represented by a blue shadow, and docking targets appear in the center
of the map view and at the edges of the application window. Each target represents an area
where the window can be positioned. The blue shadow displays where the attribute table will
be docked when you release the click.

j Pause over the right docking target, and then release the click to dock the attribute table
to the right of the map and above the histogram.

k At the bottom of the attribute table window, click the Show Selected Records button .

Copyright © 2020 Esri. All rights reserved. 7

 Spatial Data Science MOOC

The attribute table shows the records for the selected counties. You can review these records
in the table to verify their voter turnout values.

l In the right tail of the histogram, drag a box around the bins with the four highest voter
turnout values, as shown in the following graphic.

Copyright © 2020 Esri. All rights reserved. 8

 Spatial Data Science MOOC

The counties with the highest voter turnout are selected in the histogram, map, and table.

m In the histogram, click any blank area to clear the selection.

n Close the attribute table and the chart window by clicking the Close button on each
window tab, as specified in the following graphic.

Copyright © 2020 Esri. All rights reserved. 9

 Spatial Data Science MOOC

You have visualized the distribution of voter turnout values and identified where the lowest
and highest values fall on the map. Next, you will use layer symbology to visualize the spatial
distribution of voter turnout across the country.

Step 3: Change layer symbology

Currently, every county in the 2016 election results layer is symbolized using the same color.
This type of symbology is referred to as single symbol. In this step, you will change the
symbology to represent the 2016 voter turnout.

a In the Contents pane, right-click CountyElections2016 and choose Symbology.

Copyright © 2020 Esri. All rights reserved. 10

 Spatial Data Science MOOC

The Symbology pane appears.

b In the Symbology pane, under Primary Symbology, click the down arrow and choose
Graduated Colors.
Graduated Colors classifies the data into different ranges based on the values of a specified
attribute field. Each class is assigned a shade of color to show the relative difference between
the feature values. To learn more about graduated colors, see ArcGIS Pro Help: Graduated
colors.
You will specify the field and color ramp so that the symbology represents ranges of voter
turnout.

c Under Graduated Colors, for Field, choose Voter_Turnout, if necessary.

d For Method, click the down arrow.

Copyright © 2020 Esri. All rights reserved. 11

 Spatial Data Science MOOC

A list describing the available classification methods appears.

Natural Breaks (Jenks) is the default classification method because it is data driven. This
means that the symbol ranges are calculated based on the data values, making it adaptable
to different types of data distributions. Because you have determined that the voter turnout
variable is normally distributed, you will use the Standard Deviation method to classify voter
turnout. To learn more about data classification methods, see ArcGIS Pro Help: Data
classification methods.

e From the list of classification methods, choose Standard Deviation.

Copyright © 2020 Esri. All rights reserved. 12

 Spatial Data Science MOOC

f Below Classes, drag the pane divider (as specified in the following graphic) down to see
the Color Scheme parameter.

g Next to Color Scheme, click the down arrow.

h In the Color Scheme window, check the Show Names box, as specified in the following
graphic.

Copyright © 2020 Esri. All rights reserved. 13

 Spatial Data Science MOOC

i Choose the Purple-Green (Continuous) color scheme.

j In the Symbology pane, below the pane divider, click the Histogram tab.

Note: You may need to widen the Symbology pane to better see the histogram.
By applying a Graduated Colors symbology, you created what is commonly referred to as a
choropleth, or thematic, map. Choropleth maps visualize low-to-high values using light-to-
dark colors. Because you are using the Standard Deviation classification method, you applied
a diverging color scheme. A diverging color scheme classifies values based on how far they

Copyright © 2020 Esri. All rights reserved. 14

 Spatial Data Science MOOC

are from the average. On the Histogram tab, you can see how the distribution of values
corresponds to the classes of color. The counties with below-average voter turnout are
represented in shades of purple, and the counties with above-average voter turnout are
represented in shades of green.

k In the Symbology pane, next to Color Scheme, click the Color Scheme Options button
and choose Apply To Fill And Outline.

l In the Contents pane, turn on the US_States layer.

Combining the Graduated Colors symbology with the state layer overlay, you can begin to
get a sense of how states compare to each other in terms of voter turnout. States like West
Virginia (WV) and Tennessee (TN) stand out as having low voter turnout, and states like
Colorado (CO) and Minnesota (MN) stand out as having high voter turnout. You will use a bar
chart to summarize and compare voter turnout by state.

m Close the Symbology pane.

n Save the project.

Copyright © 2020 Esri. All rights reserved. 15

 Spatial Data Science MOOC

Step 4: Create a bar chart

ArcGIS Pro provides numerous ways to visualize your data in different charts. You will now
visualize the data as a bar chart.

a In the Contents pane, right-click CountyElections2016, point to Create Chart, and view
the available options.

b Choose Bar Chart.

c In the Chart Properties pane, provide the following parameters:

• Category Or Date: State

• Aggregation: Mean
• Numeric Fields: Voter_Turnout

Copyright © 2020 Esri. All rights reserved. 16

 Spatial Data Science MOOC

d Click Apply.

Copyright © 2020 Esri. All rights reserved. 17

 Spatial Data Science MOOC

The bar chart summarizes county voter turnout values by state. Each bar represents a state,
and the height of the bar corresponds to the mean voter turnout value. For more information
about bar charts, see ArcGIS Pro Help: Bar chart.

e At the bottom of the Chart Properties pane, under Sort, click the down arrow and choose
Y-Axis Descending.

Copyright © 2020 Esri. All rights reserved. 18

 Spatial Data Science MOOC

Sorting the bars by value makes it easier to visually rank the states from highest to lowest
voter turnout.

f In the chart window, select the three states with the lowest average voter turnout.

Hawaii, West Virginia, and Tennessee have the lowest average voter turnout, which confirms
what you observed in the map. The bar chart summarizes the voter turnout for each state into
a single average value. Within each state, however, there can be quite a bit of variation in
voter turnout. To examine the individual county voter turnout within each state, you can use a
filtered bar chart.

g In the chart window, clear the selection.

Step 5: Filter a chart using a selection

a Create a second bar chart for the CountyElections2016 layer using the following
parameters:

• Category Or Date: County

• Aggregation: None
• Numeric Fields: Voter_Turnout

Copyright © 2020 Esri. All rights reserved. 19

 Spatial Data Science MOOC

b Click Apply.

c At the bottom of the Chart Properties pane, under Sort, click the down arrow and choose
Y-Axis Descending.

d In the chart window, on the toolbar next to Filter, click the Filter By Selection button .
Filter By Selection filters the chart to only show selected features. Because no features are
selected yet, the bar chart is empty.

e Dock the state bar chart window to the right of the map, above the county bar chart
window.

You now have a bar chart visualizing average voter turnout by state and a bar chart visualizing
individual county voter turnout values of selected features.

f In the state bar chart, select Georgia.

Hint: Georgia is located between New Mexico and Oklahoma in the chart. Point to the bar to
see which state it represent. You can also use the Zoom Mode button to zoom in to the
bar chart.

Copyright © 2020 Esri. All rights reserved. 20

 Spatial Data Science MOOC

The county bar chart populates to show the individual county values within Georgia. Each bar
in the county bar chart corresponds to a single feature on the map, so the colors of the bars
match the map symbology.
You can use this interactive selection to see the range of individual county values within each
state. To compare the county values to the national average voter turnout value of 0.59
(identified in the histogram), you will add a guide to your chart.

g Click the county bar chart window tab to activate that chart.
The Chart Properties pane updates to show county information.

h At the top of the Chart Properties pane, click the Guides tab.

i Under Guides, click Add Guide.

Copyright © 2020 Esri. All rights reserved. 21

 Spatial Data Science MOOC

j For Value, type 0.59.

k For Line Color, click the line and choose a bright blue color.

l For Label, type National Average.

A line appears in the county bar chart marking the national average voter turnout value.

Guides allow you to reference or highlight significant values or thresholds in your charts.

m In the state bar chart, click other states to see how their county voter turnout values vary
within the state and to compare the state's average voter turnout to the national average.
Note: If you previously used the Zoom Mode button , click the Select Interaction Mode
button so that you can select other states.

n Clear the selection and close both bar charts.

Copyright © 2020 Esri. All rights reserved. 22

 Spatial Data Science MOOC

You have used bar charts and Filter By Selection to explore state voter turnout averages and
to examine the individual county voter turnout values for each state.

Step 6: Create a box plot

You can use interactive selection to see an overview of state averages and to investigate the
range of county values within individual states. To visualize and compare the distribution of
voter turnout values for every state at one time, you will create a box plot.

a Create a box plot for the CountyElections2016 layer using the following parameters:

• Numeric Fields: Voter_Turnout

• Category: State
• Show Outliers: Checked

b Click Apply.

The box plot chart allows you to visualize and compare the entire distribution of county voter
turnout values for each state. Box plots split numeric values into four equal quartiles and
visualize five key statistics for each distribution: minimum, first quartile, median, third quartile,
and maximum. The whiskers extending from the boxes span from the minimum value to the
maximum value, illustrating the full range of values found in each state. The boxes span from

Copyright © 2020 Esri. All rights reserved. 23

 Spatial Data Science MOOC

the first quartile to the third quartile, illustrating the range of the middle half of values, or the
interquartile range (IQR). The IQR indicates the size of spread, or variability, in voter turnout
values in each state. For more information about box plots, see ArcGIS Pro Help: Box plot.

c In the box plot chart, point to Texas.

The ToolTip displays the key voter turnout statistics for the state. Texas has a relatively low
voter turnout average as a state. However, there is a wide range of county voter turnout
values, spanning from approximately 0.32 to approximately 0.88. The counties with voter
turnout values that are very different from the state average are considered outliers and are
displayed as dots beyond the plot's whiskers.

d In the box plot chart, select the Texas outliers, as shown in the following graphic.

Copyright © 2020 Esri. All rights reserved. 24

 Spatial Data Science MOOC

The outliers are selected on the map.

Reviewing the overall distribution of voter turnout values in conjunction with individual feature
locations can help you understand the data and identify areas that you may want to further
investigate.

e Clear the selection and close the box plot chart.

Copyright © 2020 Esri. All rights reserved. 25

 Spatial Data Science MOOC

Step 7: Explore variable relationships in a scatter plot matrix

You have used various spatial and nonspatial data visualization techniques to explore voter
turnout values and distributions. You can also use data visualization tools to explore
relationships in your data. Because you want to predict voter turnout, you will explore the
relationship between voter turnout and other variables in your data.

a Create a scatter plot matrix for the CountyElections2016 layer.

b In the Chart Properties pane, under Numeric Fields, check the box for the following fields:

• Voter_Turnout
• 2019 Median Age
• 2019 Per Capita Income
• Own A Selfie Stick : Percent
• 2019 Education: High School/No Diploma : Percent (Note: Scroll down to the second
group of education field options.)
• Buying American Is Not Important To Me : Percent

c Check the Show Histograms box, and then click Apply.

Copyright © 2020 Esri. All rights reserved. 26

 Spatial Data Science MOOC

A scatter plot matrix is a grid of scatter plots, also referred to as mini-plots, used to visualize
bivariate relationships between combinations of variables. Each scatter plot in the matrix
visualizes the relationship between a pair of variables, allowing many relationships to be
explored in one chart. A histogram visualizing the distribution of each individual variable can
also be included in the matrix. For more information about scatter plot matrices, see ArcGIS
Pro Help: Scatter plot matrix.

d In the Chart Properties pane, check the Show Linear Trend box.

A linear trend line is added to each scatter plot in the matrix. The direction of the trend line
indicates whether the variables have a positive or negative relationship, and the R-squared
(R2) value indicates the strength of the relationship. For more information about scatter plots,
see ArcGIS Pro Help: Scatter plot.

e In the Chart Properties pane, check the Show As R2 box.

Copyright © 2020 Esri. All rights reserved. 27

 Spatial Data Science MOOC

The mini-plots in the matrix are now visualized with a color gradient corresponding to the
strength of the R-squared value. You can select any mini-plot to view the relationship in more
detail using the larger preview plot. While every pairwise combination of variables is plotted
in the matrix, you are specifically interested in how each variable relates to voter turnout. The
column of mini-plots on the far left includes the relationships between voter turnout and the
other variables.

f In the scatter plot matrix, select the mini-plot comparing Voter_Turnout and 2019 Median
Age.

Median age has a positive relationship with voter turnout, where a higher median age
corresponds to a higher voter turnout. However, the R-squared value for this trend is 0.15,

Copyright © 2020 Esri. All rights reserved. 28

 Spatial Data Science MOOC

which means that median age alone can only explain about 15 percent of the variability in the
voter turnout values.

g Select the mini-plot comparing Voter_Turnout and 2019 Per Capita Income.

Per capita income also has a positive relationship with voter turnout, where a higher per
capita income corresponds to a higher voter turnout. The R-squared value for this trend is
0.34, which means that per capita income can explain about 34 percent of the variability in
voter turnout values. Within the preview plot, you can see where some of the points deviate
from the trend. You can investigate those points using a selection.

h In the preview plot, select points that deviate from the trend to see where they fall on the
map, as shown in the following graphic.

Copyright © 2020 Esri. All rights reserved. 29

 Spatial Data Science MOOC

The selected counties are highlighted on the map. To see if the variable relationships vary
spatially, you will filter the chart by the map extent.

i Clear the selection, and then save the project.

Step 8: Explore relationships at different scales

To investigate if the strength of the variable relationships vary from place to place, you can
filter the scatter plot matrix to include only counties that are visible on the map.

Copyright © 2020 Esri. All rights reserved. 30

 Spatial Data Science MOOC

a In the scatter plot matrix window, on the toolbar next to Filter, click the Filter By Extent
button .

b From the Map tab, in the Navigate group, click Bookmarks and choose the WV, VA, MD
bookmark.

Note: The R-squared values will vary based on the size of your map and chart windows.
The chart updates to calculate the relationships between the variables of the counties visible
in the map extent.
If you compare the R-squared values at a national scale to this local scale, you can see that the
relationships between voter turnout and per capita income and voter turnout and owning a

Copyright © 2020 Esri. All rights reserved. 31

 Spatial Data Science MOOC

selfie stick has increased. However, the relationship between voter turnout and median age
has disappeared.

c Zoom and pan around the map to explore how variable relationships vary by scale and
location.
The changes in R-squared values indicate that the linear relationships between the variables
vary spatially. In the next step, you will explore and quantify different types of local
relationships using the Local Bivariate Relationships tool.

d Close the scatter plot matrix window and the Chart Properties pane.

e In the Contents pane, uncheck CountyElections2016 to turn the layer off.

Copyright © 2020 Esri. All rights reserved. 32

 Spatial Data Science MOOC

Step 9: Explore local bivariate relationships

The Local Bivariate Relationships tool quantifies the local relationship between two variables
and indicates how the type of relationship varies spatially. You will use this tool to quantify the
relationship between voter turnout and per capita income and determine if this relationship
varies across the contiguous United States. To learn more about the Local Bivariate
Relationships tool, see ArcGIS Pro Help: How Local Bivariate Relationships works.

a Zoom to the United States bookmark.

b In the Geoprocessing pane, search for local bivariate relationships.

Hint: If you closed the Geoprocessing pane: Analysis tab > Geoprocessing group > Tools.

c Open the Local Bivariate Relationships (Spatial Statistics Tools) tool and set the following
parameters:

• Input Features: CountyElections2016

• Dependant Variable: Voter_Turnout
• Explanatory Variable: 2019 Per Capita Income
• Output Features: VoterTurnout_PerCapitaIncome
• Enable Local Scatterplot Pop-ups: Checked

Copyright © 2020 Esri. All rights reserved. 33

 Spatial Data Science MOOC

d Leave all other default parameters and click Run.

e In the Contents pane, drag the US_States layer above the VoterTurnout_PerCapitaIncome
layer.

Copyright © 2020 Esri. All rights reserved. 34

 Spatial Data Science MOOC

The Local Bivariate Relationships tool identifies not only linear relationships but also concave/
convex and other undefined complex relationships. The colors in the map correspond to the
type of relationship found in that area. Based on this output, the relationship between per
capita income and voter turnout varies by location. In some areas, there is no statistically
significant relationship between the two variables. In most areas, however, there is a
statistically significant positive linear relationship to voter turnout, where voter turnout
increases as per capita income increases.

f Click one of the counties symbolized as Positive Linear to open a pop-up window
visualizing the relationship.

g Scroll down in the pop-up window or expand it until you see the scatter plot.

Copyright © 2020 Esri. All rights reserved. 35

 Spatial Data Science MOOC

You can use the pop-up windows to review a scatter plot of the selected county and its
neighbors, indicating the strength and type of the local variable relationship.

h When you are finished reviewing the information, close the pop-up window.

i Click a county symbolized as Convex to examine the shape of the relationship in the
scatter plot.

Copyright © 2020 Esri. All rights reserved. 36

 Spatial Data Science MOOC

You have verified that voter turnout is related to per capita income across the majority of the
country. Based on this information, per capita income will likely be useful in your prediction
model. To see how the relationship between voter turnout and education varies spatially, you
will run Local Bivariate Relationships again.

j Close the pop-up window.

k From the Geoprocessing pane, in the Local Bivariate Relationships tool, set the following
parameters:

• Input Features: CountyElections2016

• Dependant Variable: Voter_Turnout
• Explanatory Variable: 2019 Education: High School Diploma : Percent
• Output Features: VoterTurnout_HSDiploma
• Enable Local Scatterplot Pop-ups: Checked

l Click Run.

m In the Contents pane, drag the US_States layer above the VoterTurnout_HSDiploma layer.

Copyright © 2020 Esri. All rights reserved. 37

 Spatial Data Science MOOC

Generally, the percent of the population with a high school diploma does not have a
statistically significant relationship with voter turnout. However, there is a statistically
significant negative linear relationship in most of Maine.

n Zoom to the Maine bookmark.

o Click one of the county polygons.

Copyright © 2020 Esri. All rights reserved. 38

 Spatial Data Science MOOC

Based on this information, this variable would likely be useful if you were to create a
prediction model of voter turnout in Maine. This example demonstrates how the scale of your
analysis—in this case, country versus state—can impact which variables are relevant.

p Close the pop-up window.

q If you would like to continue with this analysis, proceed to the optional stretch goal;
otherwise, close the map and the Geoprocessing pane, and then save the project and exit
ArcGIS Pro.

Copyright © 2020 Esri. All rights reserved. 39

 Spatial Data Science MOOC

Stretch goal (Optional)

If you would like to continue analyzing variable relationships, you can continue running the
Local Bivariate Relationships tool on other variables in the dataset.
Use the following high-level tasks to continue this analysis:

1. Run the Local Bivariate Relationships tool using other variables.

2. Share your results in the Lesson Forum, using the #stretch hashtag in the posting title.
3. In the Lesson Forum, identify two to three variables that you want to use in the prediction
model.

Use the Lesson Forum to post your questions, observations, and syntax examples. Be sure to
include the #stretch hashtag in the posting title.
When you are finished, close the map and Geoprocessing pane, and then save the project
and exit ArcGIS Pro.

Copyright © 2020 Esri. All rights reserved. 40