Transcript Presented by Katherine Fraser - Triad Software Technology

VISUALIZING DATA USING
MICROSOFT POWER VIEW
Presented by
Katherine Fraser
PRESENTATION OVERVIEW
I.
What is data visualization?
II. Examples of data visualization
III. Single variable vs. multivariate data
IV. Types of data visualizations
V. Tools used to visualize information
VI. Demos using Excel Power View
DATA VISUALIZATION
Data Visualization is the effort to make information easily
perceptible by humans, enabling insight.
 Half of the human brain is devoted to processing visual information.
 Information Design: the practice of presenting information in a way
that fosters efficient and effective understanding of it, specifically for
graphic design for displaying information effectively (Wikipedia)
 Using pictures, symbols, colors, and words to communicate ideas,
illustrate information or express relationships visually.... add seeing
to reading to make complex data easier to understand (John
Emerson, backspace.com)
 Edward Tufte: concepts of graphic excellence and chartjunk
EDWARD TUFTE: SIX FUNDAMENTAL
PRINCIPLES OF ANALYTICAL DESIGN
1.
2.
3.
4.
5.
6.
Show quantitative comparisons
Show causality/explanation by placing the data in an
appropriate context (not time series)
Use multivariate analysis
Integrate evidence: words, numbers, images, diagrams
Document your source to provide credibility
Have content. Analytical presentations stand or fall based on
their content.
BAR CHARTS: SINGLE VARIABLE DATA
SCATTER PLOT: MULTIVARIATE DATA
TYPES OF DATA VISUALIZATION
 Charts
 Bar charts and column charts
 Line charts
 Scatter plot
 Sparklines
 Pie charts
 To use or not to use?
 Data-ink ratio
 Small multiples
 Maps
 Treemaps
BAR CHARTS
 Bar charts can be vertical or horizontal,
may be stacked
 Graphics should tend toward the horizontal,
i.e., be greater in length than in height;
Left-to-right comprehension
 Horizon analogy
 Ease of labeling
 Emphasis on causal influence
 Bullet graphs (vs. gauges)
BAR CHARTS
Stephanie Evergreen & Gavin McMahon, http://makeapowerfulpoint.com/
HORIZONTAL STACKED BAR CHART
CLUSTERED BAR CHARTS
Stephanie Evergreen & Gavin McMahon, http://makeapowerfulpoint.com/
LINE CHARTS: COMPARING DATA
EDWARD TUFTE: GRAPHICAL EXCELLENCE
 Consists of complex ideas communicated with clarity, precision, and
efficiency. “Graphical elegance is often found in simplicity of design
and complexity of data.”
 Gives the viewer the greatest number of ideas in the shortest time
with the least ink in the smallest space (data -ink ratio)
 Is nearly always multivariate
 Requires telling the truth about the data
NON-DATA INK
SMALL, NON-COMPARATIVE DATA
Data Availability Policies:
What do authors have to provide?
Descriptions
65.5%
Programs
62.1%
Code
51.7%
Datasets
89.7%
CHARTJUNK
EDWARD TUFTE: SPARKLINES
 Small, high-resolution graphics usually embedded in a
full context of words, numbers, images.
 Shows recent changes in relation to many past
changes (context) and reduces recency bias
EXCEL 2013
New Features
 Quick Analysis—options for analyzing data
such as totals and sparklines
 Recommended charts—subset of chart
types appropriate to the data selected
 Chart Tools—Design and Layout tabs
 Pivot tables—good for aggregations
 Power View—data visualization tool
PIE CHARTS
 Tufte, “Given their low data-density and failure to order numbers
along a visual dimension, pie charts should never be used.”
 Jen Underwood, “Most often, pie charts are misused to
communicate part-to-whole scenarios where line or bar charts
would be much more effective.”
 Pie charts are intended to display proportions of a whole within
a single, small data set. Although humans are good at
comparing linear distances along a scale—like bar graphs—pie
charts don’t bring those skills to bear. We tend to underestimate
acute angles, overestimate obtuse angles, and take horizontally
bisected angles as much larger than their vertical counterparts.
ANALOG VS. DIGITAL
 An analog visual image is easier to process rapidly than is a
number; one is mentally processed as an image and the other as
text.
 The basic rule is that a digital display works best when a value with
high precision is required, while analog works best when rate -ofchange or relationship to a limit is required.
 Work with control-panel operations: people who had to read digital
gauges had a harder time keeping a clear image of the overall
situation. They knew the individual values, but had a much lower
sense of how the overall system was performing.
 A good design must minimize mental transformation or calculations,
such as calculating how close a reading is to the high or low value.
Taken from Professional Writing course “Technical Editing and Production”, Michael J. Albers, East Carolina
University
HOW MUCH TIME HAS ELAPSED
NOT A PIE CHART
EDWARD TUFTE: SMALL MULTIPLES
 Same graphical structure repeated
 Inherently multivariate and inevitably comparative
 Constancy of design helps user focus on changes in
data
 Do work adjacent in space not serial in time (spread
over multiple pages)
MAPS
 Use geographic data to compare a variable across a map
 Examples: unemployment rate by state or the number of persons
on the various floors of a building
 A choropleth map has
shaded or patterned areas
in proportion to the
measurement of the
statistical variable being
displayed, such as
population density or percapita income.
TREEMAPS
 Invented by researcher Ben Shneiderman in 1991
 Multiple boxes concentrically nested inside of each other
 The area of a given box represents the quantity it represents
 A treemap is a compact and intuitive interface for mapping an
entity and its constituent parts
INFOGRAPHICS
 Data visualizations require more work by a user in order to
find patterns and insight; more complex and involve analysis.
 Infographics are a quick and popular way of communicating
that insight; fast, timely, with the aim of presenting
information rather than analyzing it too deeply.
 Three parts of all infographics
1.
2.
3.
The visual consists of colors and graphics. “Theme” graphics
represent the data and “reference” graphics point to additional data.
Statistics and facts usually serve as the content for infographics.
Infographics should provide insight into the data that they are
presenting.
 Tools: Visual.ly, Photoshop
INFOGRAPHICS
 Examples:
 Food Safety
 Fracking
 Healthcare literacy
 Weather forecast
DATA VISUALIZATION TOOLS
Formatted, printable reports vs. ad hoc, data discovery tools
 Static reports: good for embedding in presentations or web pages
 Miscrosoft SSRS (Reporting Services)
 Crystal Reports
 High Charts
 Data discovery: fewer formatting options, allows for on -the-fly data
analysis
 Tableau
 Excel Power View
 SAP Lumira
 D3 (open source) http://d3js.org/
 Why get data visualization software? If you already have an OLAP or
Big Data source, or a managed BI data source and you need a
specialized tool or if you have data simple enough to analyze directly
MICROSOFT POWER BI
Features delivered via Excel
 Power Query
 “Data Discovery and Access”
 SSIS (ETL tool)
 Power Pivot
 “Modeling and Analysis”
 SSAS (in-memory storage)
 Power View and Power Map
 “Visualization”
 SSRS
 Power View requires ProPlus version
EXCEL POWER VIEW: DEMO
Excel demo
 Power Query to import data
 PowerPivot to store data
 Power View to visualize data
 Tiles/Cards
 Scatterplot
EXCEL POWER VIEW: DEMO
Excel demo
 Power View to visualize data
 Matrix view, small multiples
 Scatterplot
 Map
DATA VISUALIZATION PEOPLE
 Edward Tufte “The Visual Display of Quantitative Information”
 David McCandless, informationisbeautiful.net
 Jer Thorp, Data Artist for NY Times, TED Talk @ blprnt
 Jen Stirrup, Microsoft MVP @jenstirrup
 Naomi Robbins “Creating More Effective Graphs”
 Alberto Cairo, theFunctionalArt.com
 Stephen Few, dashboard design
 Nathan Yau, flowingdata.com
 datajournalismhandbook.org
 vizualize.tumblr.com
THANK YOU!
Katherine Fraser
 [email protected]
 @sqlsassy