Visual Data Insights Using SAS ODS Graphics: A Guide to Communication-Effective Data Visualization

✍ Scribed by LeRoy Bessler

• Publisher: Apress
• Tongue: English
• Leaves: 627
• Category: Library

✦ Synopsis

SAS ODS graphics users will learn in this book how to visually understand and communicate the significance of data to deliver images for quick and easy insight, with precise numbers.

Many charts or plots require the viewer to run the eye from a bar end or plot point to some point on an axis, and then to interpolate between tick marks to estimate the value. Some design choices can lead to wrong conclusions or mistaken impressions. Graphic software relies on defaults to deliver something if you make a minimal effort, but that something is not likely to be exactly what you want.

Visual Data Insights Using SAS ODS Graphics provides examples using experience-based design principles. It presents examples of bar charts, pie charts, and trend lines or time series plots, the graph types commonly used in business, other organizations, and the media for visual insight into data. Newer graphs are also included: dot plots, needle plots, waterfall charts, butterfly charts, heat maps, bubble plots, step plots, high-low plots, and donut charts. In addition, there are basic tools of statistics: scatter plots, box plots, histograms, fit and confidence plots, and distributions.

Author LeRoy Bessler introduces unique creations, including sparsely annotated time series, maximally informative bar charts, better box plots, histograms based on interesting atypical rationales, and much more.

The examples use SAS sample data sets as input. Any SAS user can experiment with the code presented to see what else is possible, or adapt it to repurpose the design and apply it with a customized version of that code.

What You’ll Learn

• Create graphs that are easily and quickly interpreted, and without ambiguity
• Supply precise data values that are correct on the graph and correctly associated with the graphic visual elements
• Take advantage of widely applicable (but not necessarily available elsewhere) design examples
• Avoid bad practices that are encouraged by poor examples elsewhere
• Get past sub-optimal designs and results that are built into software defaults
• Take advantage of less familiar capabilities available in the software

Who This Book Is For

SAS software users who want to understand their data and/or visually deliver their results

✦ Table of Contents

Contents
About the Author
About the Technical Reviewer
Acknowledgments
Introduction
Part I: Design Principles
Chapter 1: Principles of Communication-Effective Graphic Design
Be Brief, Clear, Picturesque, and Accurate
3D Pie Charts Are Always Misleading
3D Bar Charts Are Anticommunicative
Graphs Need Image and Precise Numbers
A Usable Stacked Bar Chart Requires an Axis Table to Deliver the Precise Values
A Clustered Horizontal Bar Chart Is Better Than a Stacked Bar Chart
For Bar Charts, Horizontal Is Usually Better
For a Line Plot with Discrete X Values, an X Axis Table Is an Alternative to Annotation
Curve Labels Eliminate the Need for a Legend
Show Them What’s Important
Show Them What’s Important with Ranking
Show Them What’s Important with Subsetting
Three Ways to Do Ranked Subsetting
Simplicity Accelerates and Facilitates Visual Insights into Data
A Sparse Graph Is Easily and Quickly Interpreted
Inform the Viewer About the Key Data Points of a Multi-line Overlay Plot
For a Trend, Usually Start the Y Axis at Zero
For a Bar Chart, Unless There Are Negative Values, Always Start the Value Axis at Zero
Use Maximally Simple Design to Focus on What’s Important
Tell Them What’s Important with a Headline
A Graph Footnote Does Not Need to Be Small
Assure Text Readability
Font Size and Font Weight Affect Readability
Keep Text Horizontal
Never Use Backgrounds—They Impair Readability
Readability Depends on Display Situation
A Graph Can Have a Companion Table
Web Graphs
Include Data Tips (a.k.a. Mouseover Text)
A Web Graph Can Have a Companion Hyperlinked Excel Table
A Web Graph Should Not Require Scrolling
Summary
Chapter 2: Principles of Communication-Effective Use of Color
Avoid Red and Green for “Bad Versus Good”
Color-Coding Data with a Multiple Shades of the Same Hue
Use Color to Communicate, Not to Decorate
Use of Color Can Confuse
Establish and Use a Personal Color Palette for Consistency
Beware of Color Names
Benefits of Boring Black and White
Color Requires Sufficient Mass to Be Distinguishable
Never Use Background Images or Textured or Color Gradient Backgrounds
Use a Plain Solid Color Background
Provide High Contrast Between Text Color and Background Color
Pie Charts and Color
Emphasis Options for Colorless Text
Choosing the Right Colors
A Light Color Can Be the Right Color
Uncolor Might Be the Right Color
“Transparent” Color As the Right Color
Color Differs on Different Media
Color Systems
RGB Colors
HLS Colors
Examples of Good Color Use
Summary
Part II: Widely Applicable Examples You Can Use
Chapter 3: Introduction to SAS ODS Graphics
Outer Structure of ODS Graphics Code in Examples
Inner Structure of ODS Graphics Code
Text Attributes Control in ODS Graphics
Measurement Units
Borders
From Defaults Through Customization for a Simple Scatter Plot
Control of Text Attributes with a Custom ODS Style
For More Introduction and Added Information
Summary
Chapter 4: Bar Charts, Butterfly Charts, Waterfall Charts, Dot Plots, Needle Plots, Area Bar Charts, Text Graphs, and Line Charts: Charts for Categorical Data
Present the Categories in Your Charts and Plots in Ranked Order
Provide Precise Numbers for Your Charts and Plots for Categorical Data
Bar Charts
Bar Chart Types
General Remarks About Code Used
Input Data and Discrete Attribute Map for the Examples
Horizontal Bar Charts
Default Horizontal Bar Chart
Easy Communication Effectiveness Improvements for a Horizontal Bar Chart
Maximally Informative Horizontal Bar Chart—Ranked by Measure of Interest
Subsetted Ranked Fully Informative Horizontal Bar Charts
Horizontal Charts for One Category Variable and Two Response Variables
Dot Plots
Vertical Bar Charts
Waterfall Chart
Needle Plots
Single-Needle Needle Plots
Review of Tree Chart, Flag Chart, and CrossRoads SignPost Chart and the Alternatives of Pie Chart, Vertical Bar Chart, and Horizontal Bar Chart
Area Bar Chart
Summary
Chapter 5: Pie Charts and Donut Charts
General Remarks About Code Used
Examples and Methods for Pie Charts
A Tour of What You Can Do with Donut Charts
What Was Really My First Donut Chart
My Latest Donut Charts
Summary
Chapter 6: Heat Maps
General Remarks About Code Used
Types of Heat Maps
Heat Maps with Character Variables for X and Y
Heat Map with Multifunction Annotation and a Categorical Response Variable
Heat Maps with a Character Variable for Y and a Numeric Variable for X
Heat Map with Character Y Values and Numeric X Values Converted to Character
Heat Maps with Both X and Y As Numeric Data
So, How Can We Get Some Insight Visually into This Data?
Summary
Chapter 7: Bubble Plots
General Remarks About Code Used
Examples and Methods
Options Available but Not Used Here
Summary
Chapter 8: Time Series Plots and Trend Lines
General Remarks About Code Used
Key Concerns and Methods for Time Series Plots
Tools for Time Series Graphs
Options for Lines and Markers
Special Data Sets Used for Some Time Series Graphs
Inspecting and Managing the SASUSER Data Library
Time Series One-Line Plots
Getting Started
Using the ODS Graphics SERIES Statement for a Single-Line Time Series Plot
The Most Efficient Time Series Plot: No Plot Line, No Markers, No Axis Lines
The Simplest, Yet Most Importantly Informative, Time Series Plot
Using a Reference Line in the Time Series Plot
Time Series Graphs Can Be Created Without the SERIES Statement
Vertical Bar Charts for Time Series Data
Needle Plot for Time Series Data
Step Plots for Time Series Data
Applying Bands to Time Series Plots
Time Series Plots Using the HIGHLOW Statement
Multi-line Time Series Plots
Overlay Plots of Two Response Variables with Different Units by Date
Multi-line Overlay Time Series Plots for Data with a Group Variable
Overlay Plots for Nearly Sparse Lines
Very Long Time Series Plot of Monthly Data
Daily Time Series Plots
One Month of Time Series Data
Single-Line Daily Time Series Plot for a Year
Multi-line Monthly Plot for Daily Time Series Data of One Year
Adding Value to the Multi-line Plot When Data Labels Are Impossible
Adding More Value to the Multi-line Plot
Seeing the Big Picture, Seeing It Better in Pieces, Seeing It Differently
Summary
Part III: Other Features
Chapter 9: Graphic Composites with PROC SGPANEL
Why Axis Offsets Are Important in PROC SGPANEL Applications
How PROC SGPANEL Can Affect Image Dimensions
General Remarks About Code Used
Input Data for the Examples
Included Code
Composites of Time Series Plots
Panels of Monthly Data
Sparse Line Tables and Sparse Line Panels
Panels for a Year of Days by Week
Composites of Categorical Data
Composites of Horizontal Bar Charts
Composites of Vertical Bar Charts
Composites of Dot Plots
Composites of Needle Plots
Composites of Scatter Plots
Composites with Three Class Variables
Summary
Chapter 10: Scatter Plots in Composites Using PROC SGSCATTER
Image Size Determination
General Remarks About Code Used
Special Data Requirement for This Chapter
PROC SGSCATTER PLOT Versus PROC SGPLOT SCATTER
Creating Composites with PROC SGSCATTER
Summary
Chapter 11: Fits and Confidence Plots
General Remarks About Code Used
ELLIPSE
Fit Statements for PROC SGPLOT (and PROC SGPANEL)
Summary
Chapter 12: Distributions, Histograms, Box Plots, and Alternative Tools
General Remarks About Code Used
Box Plots and the HBOX and VBOX Statements
The Basic Box Plot
Building an Informative Box Plot
Comparing the Informative Box Plot, with and Without Outliers
Effective Use of the Basic Box Plot
Histograms
The Basic Histogram
A Histogram to Look Deeper
The Ultimate Histogram: A Bin for Each (Integer) Value
The Significance-Based Histogram: Binning Based on a Rationale
The Normality Test Histogram: Seven Bins for Values Within One, Two, and Three Standard Deviations and the Minimum and Maximum
The Quantiles Histogram: N Bins with All the Same Number of Observations
How to Display the Actual Distribution
Image and Precise Numbers: Showing the Actual Distribution and Its Statistics—with Their Locations
Density Plots
The Normal Distribution and the PDF Function
The Normal Distribution As an Overlay to the Real Distribution
Putting It All Together: Frequency Distribution, Normal Distribution, and Statistics
An Alternative to the VBOX (Vertical Box) Plot
Summary
Chapter 13: Creating Composites of Graphs, Tables, and Text with ODS LAYOUT
Progressive Customization of a Table Created As an Image with ODS PRINTER
Direct Customization Without Customized Styles
Borders for Tables As Images: Using a Tool That Will Apply a Border to Any Image
General-Purpose Macro to Create a Table As an Image
Building a Composite with ODS LAYOUT and ODS PRINTER
Capturing ODS Objects from SAS Procedure Output As Images
Summary
Chapter 14: Delivering Precise Numbers and Alternative Views for Graphs Using SAS ODS HTML5
ODS HTML5 Code Basics
Custom ODS Styles Used in This Chapter
Time Series Data Used in This Chapter
Examples
More Complex Composites, Using Gridded Versus Absolute Layouts
Drill-Down Links to Other Web Pages from Elements of a Web Graph
Control of the Web Page Background Color and Graph Transparency (Pink Is Not Recommended)
The Web-Enabled Sparse Line Table
Summary
Chapter 15: Delivering Precise Numbers When Using PROC SGMAP
Summary
Appendix A
About These Three Images
For Further Information on SAS ODS Graphics
Index