Chapter 2

Organizing and
Visualizing Variables
 Objectives
In this chapter you learn:
◼ How to organize and visualize categorical
variables.
◼ How to organize and visualize numerical
variables.
◼ How to summarize a mix of variables.
◼ How to avoid making common errors when
organizing and visualizing variables.
 Organizing Data Creates Both
Tabular And Visual Summaries
DCOVA
◼ Summaries both guide further exploration and
sometimes facilitate decision making.

◼ Visual summaries enable rapid review of larger

amounts of data & show possible significant
patterns.

◼ Often, the Organize and Visualize step in

DCOVA occur concurrently.
Categorical Data Are Organized By
Utilizing Tables
DCOVA
Categorical
Data

Tallying Data

One Two
Categorical Categorical
Variable Variables

Summary Contingency
Table Table
 Organizing Categorical Data:
Summary Table DCOVA
▪ A summary table tallies the frequencies or percentages of items in a set
of categories so that you can see differences between categories.

Devices Millennials Use to Watch Movies or Television Shows

Devices Used To Watch Movies or TV Shows Percent

Television Set 49%
Tablet 9%
Smartphone 10%
Laptop / Desktop 32%

Source: Data extracted and adapted from A. Sharma, “Big Media Needs to Embrace
Digital Shift Not Fight It,” Wall Street Journal, June 22, 2016, p. 1-2.
 A Contingency Table Helps Organize
Two or More Categorical Variables
DCOVA
◼ Used to study patterns that may exist between
the responses of two or more categorical
variables.

◼ Cross tabulates or tallies jointly the responses

of the categorical variables.

◼ For two variables the tallies for one variable are

located in the rows and the tallies for the
second variable are located in the columns.
 Contingency Table - Example
DCOVA
◼ A random sample of 400
Contingency Table Showing
invoices is drawn. Frequency of Invoices Categorized
◼ Each invoice is categorized By Size and The Presence Of Errors
as a small, medium, or large No
amount. Errors Errors Total
Small 170 20 190
◼ Each invoice is also Amount
examined to identify if there Medium 100 40 140
are any errors. Amount
◼ This data are then organized Large 65 5 70
in the contingency table to Amount
the right. Total 335 65 400
 Contingency Table Based On
Percentage Of Overall Total DCOVA
No
Errors Errors Total 42.50% = 170 / 400
Small 170 20 190 25.00% = 100 / 400
Amount 16.25% = 65 / 400
Medium 100 40 140
Amount No
Large 65 5 70 Errors Errors Total
Amount Small 42.50% 5.00% 47.50%
Total 335 65 400 Amount
Medium 25.00% 10.00% 35.00%
Amount
83.75% of sampled invoices
Large 16.25% 1.25% 17.50%
have no errors and 47.50% Amount
of sampled invoices are for Total 83.75% 16.25% 100.0%
small amounts.
 Contingency Table Based On
Percentage of Row Totals DCOVA
No
Errors Errors Total 89.47% = 170 / 190
Small 170 20 190 71.43% = 100 / 140
Amount 92.86% = 65 / 70
Medium 100 40 140
Amount No
Large 65 5 70 Errors Errors Total
Amount Small 89.47% 10.53% 100.0%
Total 335 65 400 Amount
Medium 71.43% 28.57% 100.0%
Amount
Medium invoices have a larger
Large 92.86% 7.14% 100.0%
chance (28.57%) of having Amount
errors than small (10.53%) or Total 83.75% 16.25% 100.0%
large (7.14%) invoices.
 Contingency Table Based On
Percentage Of Column TotalsDCOVA
No
Errors Errors Total 50.75% = 170 / 335
Small 170 20 190 30.77% = 20 / 65
Amount
Medium 100 40 140
Amount No
Large 65 5 70 Errors Errors Total
Amount Small 50.75% 30.77% 47.50%
Total 335 65 400 Amount
Medium 29.85% 61.54% 35.00%
Amount
There is a 61.54% chance
Large 19.40% 7.69% 17.50%
that invoices with errors are Amount
of medium size. Total 100.0% 100.0% 100.0%
 Tables Used For Organizing
Numerical Data
DCOVA
Numerical Data

Ordered Array Frequency Cumulative

Distributions Distributions
 Organizing Numerical Data:
Ordered Array
DCOVA
▪ An ordered array is a sequence of data, in rank order, from the
smallest value to the largest value.
▪ Shows range (minimum value to maximum value).
▪ May help identify outliers (unusual observations).

Age of Day Students

Surveyed
16 17 17 18 18 18
College
Students 19 19 20 20 21 22
22 25 27 32 38 42
Night Students
18 18 19 19 20 21
23 28 32 33 41 45
 Organizing Numerical Data:
Frequency Distribution
DCOVA
▪ The frequency distribution is a summary table in which the data are
arranged into numerically ordered classes.

▪ You must give attention to selecting the appropriate number of class

groupings for the table, determining a suitable width of a class grouping,
and establishing the boundaries of each class grouping to avoid
overlapping.

▪ The number of classes depends on the number of values in the data. With
a larger number of values, typically there are more classes. In general, a
frequency distribution should have at least 5 but no more than 15 classes.

▪ To determine the width of a class interval, you divide the range (Highest
value–Lowest value) of the data by the number of class groupings desired.
 Organizing Numerical Data:
Frequency Distribution Example
DCOVA

Example: A manufacturer of insulation randomly

selects 20 winter days and records the daily high
temperature.

24, 35, 17, 21, 24, 37, 26, 46, 58, 30, 32, 13, 12, 38, 41, 43, 44, 27, 53, 27
 Organizing Numerical Data:
Frequency Distribution Example
DCOVA
▪ Sort raw data in ascending order:
12, 13, 17, 21, 24, 24, 26, 27, 27, 30, 32, 35, 37, 38, 41, 43, 44, 46, 53, 58.
▪ Find range: 58 - 12 = 46.
▪ Select number of classes: 5 (usually between 5 and 15).
▪ Compute class interval (width): 10 (46/5 then round up).
▪ Determine class boundaries (limits):
▪ Class 1: 10 but less than 20.
▪ Class 2: 20 but less than 30.
▪ Class 3: 30 but less than 40.
▪ Class 4: 40 but less than 50.
▪ Class 5: 50 but less than 60.
▪ Compute class midpoints: 15, 25, 35, 45, 55.
▪ Count observations & assign to classes.
 Organizing Numerical Data: Frequency
Distribution Example
DCOVA
Data in ordered array:
12, 13, 17, 21, 24, 24, 26, 27, 27, 30, 32, 35, 37, 38, 41, 43, 44, 46, 53, 58

Class Midpoints Frequency

10 but less than 20 15 3

20 but less than 30 25 6
30 but less than 40 35 5
40 but less than 50 45 4
50 but less than 60 55 2
Total 20
 Organizing Numerical Data: Relative &
Percent Frequency Distribution Example
DCOVA
Relative
Class Frequency Percentage
Frequency
10 but less than 20 3 .15 15%
20 but less than 30 6 .30 30%
30 but less than 40 5 .25 25%
40 but less than 50 4 .20 20%
50 but less than 60 2 .10 10%
Total 20 1.00 100%

Relative Frequency = Frequency / Total, e.g. 0.10 = 2 / 20

 Organizing Numerical Data: Cumulative
Frequency Distribution Example
DCOVA
Cumulative Cumulative
Class Frequency Percentage
Frequency Percentage

10 but less than 20 3 15% 3 15%

20 but less than 30 6 30% 9 45%
30 but less than 40 5 25% 14 70%
40 but less than 50 4 20% 18 90%
50 but less than 60 2 10% 20 100%
Total 20 100% 20 100%

Cumulative Percentage = Cumulative Frequency / Total * 100 e.g. 45% = 100*9/20

Why Use a Frequency Distribution?
DCOVA
◼ It condenses the raw data into a more
useful form.
◼ It allows for a quick visual interpretation of
the data.
◼ It enables the determination of the major
characteristics of the data set including
where the data are concentrated /
clustered.
 Frequency Distributions:
Some Tips
DCOVA
◼ Different class boundaries may provide different pictures for
the same data (especially for smaller data sets).

◼ Shifts in data concentration may show up when different

class boundaries are chosen.

◼ As the size of the data set increases, the impact of

alterations in the selection of class boundaries is greatly
reduced.

◼ When comparing two or more groups with different sample

sizes, you must use either a relative frequency or a
percentage distribution.
 Visualizing Categorical Data
Through Graphical Displays
DCOVA
Categorical
Data
Visualizing Data

Summary Contingency
Table For One Table For Two
Variable Variables

Bar Pareto Side By Side Doughnut

Chart Chart Bar Chart Chart

Pie or
Doughnut Chart
 Visualizing Categorical Data:
The Bar Chart
DCOVA
▪ The bar chart visualizes a categorical variable as a series of bars. The
length of each bar represents either the frequency or percentage of
values for each category. Each bar is separated by a space called a gap.

Devices Percent
Used to
Watch
Television Set 49%
Tablet 9%
Smartphone 10%
Laptop / 32%
Desktop
 Visualizing Categorical Data:
The Pie Chart
DCOVA
▪ The pie chart is a circle broken up into slices that represent categories.
The size of each slice of the pie varies according to the percentage in
each category.

Devices Percent
Used to
Watch
Television Set 49%
Tablet 9%
Smartphone 10%
Laptop / 32%
Desktop
 Visualizing Categorical Data:
The Doughnut Chart DCOVA
▪ The doughnut chart is the outer part of a circle broken up into pieces
that represent categories. The size of each piece of the doughnut varies
according to the percentage in each category.

Devices Percent
Used to
Watch
Television Set 49%
Tablet 9%
Smartphone 10%
Laptop / 32%
Desktop
 Visualizing Categorical Data:
The Pareto Chart
DCOVA
◼ Used to portray categorical data (nominal
scale).
◼ A vertical bar chart, where categories are
shown in descending order of frequency.
◼ A cumulative polygon is shown in the same
graph.
◼ Used to separate the “vital few” from the “trivial
many.”
 Visualizing Categorical Data:
The Pareto Chart (con’t) DCOVA
Ordered Summary Table For Causes
Of Incomplete ATM Transactions
Cumulative
Cause Frequency Percent Percent
Warped card jammed 365 50.41% 50.41%
Card unreadable 234 32.32% 82.73%
ATM malfunctions 32 4.42% 87.15%
ATM out of cash 28 3.87% 91.02%
Invalid amount requested 23 3.18% 94.20%
Wrong keystroke 23 3.18% 97.38%
Lack of funds in account 19 2.62% 100.00%
Total 724 100.00%
Source: Data extracted from A. Bhalla, “Don’t Misuse the Pareto Principle,” Six Sigma Forum
Magazine, May 2009, pp. 15–18.
 Visualizing Categorical Data:
The Pareto Chart (con’t) DCOVA

The “Vital
Few”
 Visualizing Categorical Data:
Side By Side Bar Charts DCOVA
▪ The side by side bar chart represents the data from a contingency table.

No
Errors Errors Total
Invoice Size Split Out By Errors
Small 50.75% 30.77% 47.50% & No Errors
Amount
Medium 29.85% 61.54% 35.00% Errors

Amount
Large 19.40% 7.69% 17.50% No Errors

Amount
0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0%
Total 100.0% 100.0% 100.0% Large Medium Small

Invoices with errors are much more likely to be of

medium size (61.5% vs 30.8% & 7.7%).
 Visualizing Categorical Data:
Doughnut Charts DCOVA
▪ A Doughnut Chart can be used to represent the data from a contingency table.

No
Invoice Size & Errors
Errors Errors Total Inner Ring With Errors, Outer Ring No Errors

Small 50.75% 30.77% 47.50% 19.4%

Amount 7.7%
30.8% 30.8%

Medium 29.85% 61.54% 35.00%

61.5%
Amount
29.9%
Large 19.40% 7.69% 17.50%
Amount
Small Medium Large

Total 100.0% 100.0% 100.0%

Invoices with errors are much more likely to be of

medium size (61.5% vs 30.8% & 7.7%).
 Visualizing Numerical Data
By Using Graphical Displays
DCOVA
Numerical Data

Frequency Distributions
Ordered Array and
Cumulative Distributions

Stem-and-Leaf
Histogram Polygon Ogive
Display
 Stem-and-Leaf Display
DCOVA

◼ A simple way to see how the data are distributed

and where concentrations of data exist.

METHOD: Separate the sorted data series

into leading digits (the stems) and
the trailing digits (the leaves).
 Organizing Numerical Data:
Stem and Leaf Display
DCOVA
▪ A stem-and-leaf display organizes data into groups (called
stems) so that the values within each group (the leaves)
branch out to the right on each row.
Age of College Students

Age of Day Students Day Students Night Students

Surveyed
16 17 17 18 18 18 Stem Leaf
College Stem Leaf
Students 19 19 20 20 21 22
1 67788899 1 8899
22 25 27 32 38 42
Night Students 2 0012257 2 0138
18 18 19 19 20 21
3 28 3 23
23 28 32 33 41 45
4 2
4 15
 Visualizing Numerical Data:
The Histogram
DCOVA

▪ A vertical bar chart of the data in a frequency distribution is

called a histogram.

▪ In a histogram there are no gaps between adjacent bars.

▪ The class boundaries (or class midpoints) are shown on the

horizontal axis.

▪ The vertical axis is either frequency, relative frequency, or

percentage.

▪ The height of the bars represent the frequency, relative

frequency, or percentage.
 Visualizing Numerical Data:
The Histogram
DCOVA
Relative
Class Frequency Percentage
Frequency

10 but less than 20 3 .15 15

20 but less than 30 6 .30 30
30 but less than 40 5 .25 25
40 but less than 50 4 .20 20
50 but less than 60 2 .10 10
8
Total 20 1.00 100
Histogram:Age
Histogram: Temperature
Of Students
6

Frequency
4
(In a percentage
histogram the vertical
axis would be defined to 2
show the percentage of
observations per class).
0
5 15 25 35 45 55 More
 Visualizing Numerical Data:
The Percentage Polygon
DCOVA

▪ A percentage polygon is formed by having the midpoint of

each class represent the data in that class and then connecting
the sequence of midpoints at their respective class
percentages.

▪ The cumulative percentage polygon, or ogive, displays the

variable of interest along the X axis, and the cumulative
percentages along the Y axis.

▪ Useful when there are two or more groups to compare.

Visualizing Numerical Data:
The Frequency Polygon DCOVA
Useful When Comparing Two or More Groups
Visualizing Numerical Data:
The Percentage Polygon
DCOVA
Visualizing Numerical Data:
The Cumulative Percentage Polygon (Ogive)
DCOVA
Useful When Comparing Two or More Groups
Visualizing Numerical Data:
The Cumulative Percentage Polygon (Ogive)
DCOVA
Visualizing Two Numerical Variables
By Using Graphical Displays
DCOVA

Two Numerical
Variables

Scatter Time-
Plot Series
Plot
 Visualizing Two Numerical
Variables: The Scatter Plot
DCOVA
▪ Scatter plots are used for numerical data consisting of paired
observations taken from two numerical variables.

▪ One variable’s values are displayed on the horizontal or X

axis and the other variable’s values are displayed on the
vertical or Y axis.

▪ Scatter plots are used to examine possible relationships

between two numerical variables.
 Scatter Plot Example
DCOVA

Volume Cost per

per day day Cost per Day vs. Production Volume
23 125
250
26 140
200
Cost per Day

29 146
150
33 160
100
38 167
50
42 170
0
50 188
20 30 40 50 60 70
55 195
Volume per Day
60 200
Visualizing Two Numerical
Variables: The Time Series Plot
DCOVA
◼ A Time-Series Plot is used to study
patterns in the values of a numeric
variable over time.

◼ The Time-Series Plot:

◼ Numeric variable’s values are on the
vertical axis and the time period is on
the horizontal axis.
 Time Series Plot Example
DCOVA
Number of
Year Franchises
2009 43
2010 54
2011 60
2012 73
2013 82
2014 95 Number of Franchises, 1996-2004

2015 107 120

Number of Franchises
2016 99 100
80
2017 95 60
40
20
0
2008 2010 2012 2014 2016 2018
Year
Organizing A Mix Of Variables: The
Multidimensional Contingency Table
DCOVA
◼ A multidimensional contingency table is constructed by
tallying the responses of three or more categorical variables.
◼ Can be used to discover possible patterns and relationships in
multidimensional data that simpler tables and charts would fail to
make apparent.

◼ As a practical rule, tables should be limited to no more than

three or four variables.

◼ In typical use, these tables:

◼ Extend contingency tables to two or more row or column variables, or
◼ Replace the frequencies found in a contingency table with summary
information about a numeric variable.
 A Multidimensional Contingency Table
Tallies Responses Of Three or More
Categorical Variables
DCOVA

Two Dimensional Table Showing Three Dimensional Table

Fund Type and Risk Level for Showing Fund Type, Market
sample of 479 retirement funds. Cap, and Risk Level for the
sample of the 479 retirement
funds.
 Excel, Minitab, and JMP Can Be Used To
Create Multidimensional Contingency Tables
DCOVA
◼ In Excel creating a Pivot Table yields an interactive
display of this type.

◼ In JMP you can create a table that is also interactive.

◼ In Minitab you can create such a table but it is not

interactive.

◼ JMP and Minitab have many specialized statistical &

graphical procedures (not covered in this book) to
analyze & visualize multidimensional data.
Drilling-Down On A Table Reveals
The Data The Table Summarizes
DCOVA
◼ Clicking a cell in an Excel table displays the
rows of data associated with that cell.

◼ Clicking a cell in a JMP table highlights those

the rows of data that are the source for that cell.

◼ Drill-down is perhaps the simplest form of data

discovery.
Drill-Down Reveals The Data
Underlying A Higher-Level Summary
DCOVA
Results of drilling down to
the details about small value
funds with low risk revealing
the ten-year return ranges from
4.83% to 9.44%.
Displays To Visualize A Mix Of Many
Variables
DCOVA
◼ Displays are more useful than a multidimensional
contingency table with many row and column
variables.
◼ The data (not just summary statistics) can be
shown for numerical variables.
◼ Multiple numerical variables can be presented in
one summarization.
◼ Visualizations can reveal patterns that can be
hard to see in tables.
Colored Scatter Plots Visualize Both
Numerical Variables & Categorical Variable(s)
Observations: DCOVA
Large Market Capitalization Funds (red dots)
1. Relatively have best returns and lowest expense ratios.

2. Some have either low returns or high expense ratios or both.

JMP Colored
Scatter Plot
 False Impressions Can Be
Created In Many Ways DCOVA

◼ Selective summarization:
◼ Presenting only part of the data collected.

◼ Improperly constructed charts:

◼ Potential pie chart issues.
◼ Improperly scaled axes.
◼ A Y axis that does not begin at the origin or is a
broken axis missing intermediate values.

◼ Chartjunk.
How Obvious Is It That Both Pie Charts
Summarize The Same Data? DCOVA

Why is it hard to tell? What would you do to improve?

 Graphical Errors:
No Relative Basis DCOVA

Bad Presentation ✓Good Presentation

A’s received by A’s received by
Freq. students. % students.
30%
300
200 20%

100 10%

0 0%
FR SO JR SR FR SO JR SR

FR = Freshmen, SO = Sophomore, JR = Junior, SR = Senior

 Graphical Errors:
Compressing the Vertical Axis
DCOVA

Bad Presentation ✓ Good Presentation

Quarterly Sales Quarterly Sales
$ $
200 50

100 25

0 0
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4
 Graphical Errors: No Zero Point
on the Vertical Axis
DCOVA

Bad Presentation
✓ Good Presentations

Monthly Sales $ Monthly Sales

$ 45
45
42
42 39
39 36
36 0
J F M A M J J F M A M J

Graphing the first six months of sales

 Graphical Errors: Chart Junk,
Can You Identify The Junk?
DCOVA

Bad Presentation ✓ Good Presentation

Left illustration adapted from S. Watterson, “Liquid Gold—Australians Are Changing the World of Wine. Even the French Seem Grateful.” Time,
November 22, 1999, p. 68-69
 Best Practices for Constructing
Visualizations DCOVA

▪ Use the simplest possible visualization.

▪ Include a title & label all axes.
▪ Include a scale for each axis if the chart contains axes.
▪ Begin the scale for a vertical axis at zero & use a
constant scale.
▪ Avoid 3D or “exploded” effects & the use of chartjunk.
▪ Use consistent colorings in charts meant to be compared.
▪ Avoid using uncommon chart types including radar,
surface, bubble, cone, and pyramid charts.
 Chapter Summary
In this chapter we covered:
◼ Organizing and visualizing categorical variables.
◼ Organizing and visualizing numerical variables.
◼ Summarizing a mix of variables.
◼ Avoiding common errors when organizing and
visualizing variables.