Greg Nottes, workshop presenter
Tools — extracts perfect info and meaning, doesn’t exist 🙂
Not here to necessarily talk about big data, but focus on data to communicate and tell a story; Understanding it. Then looking for errors. Or asking question about anomalies.
Why use different types of viz tools?
Word clouds — frequency of words are visually represented. tagul.com (?) not right…..
Great for decoration — but not powerful.
How do you get meaning from data? Pull different viewpoints
Word cloud vs this viewpoint — not much can be told at a glance with this one; tells those who are invested in the data what’s going on, but not those outside the audience.
Remember you’re the closest to the data, you know it best; what are the key points to communicate to your audience. How much time your audience will take with this data.
One second story vs. looking more closely.
- Check the data
- Explain encodings (clear legends)
- Label Axes
- Include Units
- Keep Geometry in Check
- Include Sources
- Consider Audience
Greg’s library, Reference data previously collected w LibStats (OSS) now using SpringShare product
In Excel, it’s really helpful to freeze top row — helps read long set of data. Also, filtering on the rows — helps quickly filter specific sets of data.
Ahead of time, setting up instructions for directly checking — making a list, transforming…
Ask lots of questions:
- How accurate is the data?
- Where is it gathered from?
- What are its weaknesses?
Question asked — could a statistician be consulted by reference librarians to find out how to get a random sample to find meaningful data?
Calculating more meaningful times (grouping data)
Pivot Tables (in Excel) — Range; build report; allows you to take data, gives sum/average for any criteria as a row/column. Step toward getting graph you want.
Every element you add to a viz, increases the cognitive load on the viewer. x axes/y axis. No title showing summary of what’s been shown.
What are the key things here you want to share with others? That’s really what matters.
- Checking the data is crucial in telling the story.
- Explain encodings; use legend; colors; . At a quick look, viewer can tell.
- Label axis — whats on the x and y axis and label them
- What are the units
- Keep geometry in check (look at something and compare) Bar chart easier to tell than pie because area in a circle harder to distinguish in a pie chart.
- Include your sources
- Consider audience critical one
Additionally, keep these in consideration:
- Data understanding
- Grouping — what makes sense for groupings? Depends on the story you’re trying to tell; results may require you to group differently…
Show one thing:
- Highlight Key Points
- Comparing Parts of a whole: Pie or stacked bar chart
Consider graphic and how to present.
Three keys to data viz:
- What’s the key message?
- What’s the time for cognitive load to get that?
- Keeping it simple
- Tableau Public & Tom Newman Worksheets for the Public Library Survey
- Good viz is planning — what are the data streams we have and how can we look at it?
- Juice Labs — see different ways to display your data; which kind of these will help tell your story.
Common Charts: Pie, Bar, Line & Other Types:
- Numerical data: chart; Choosing best type
- Component comparison — Percentage of a total; share percentage of total; accounted for X percent — pie chart; adding labels really help. Pie charts work best with two or three; not much more
- Item comparison, ranking of items:
- larger than
- smaller than
- equal to
- rank — bar charts
- Time series comparison: changes over time — column or line chart;
- Frequency distribution: items within ranges; range of concentration x to y range; step or line chart
- Correlation comparison: relationship bw variables; related to, increases with, changing, decreasing with — scatter/dot/bubble chart; paired bar char
- Data maps (geographic data)
- No numerical data — pushpin; multiple symbol
- Numerical (single field) — shared area; shaded/sized circle; multiple symbol
- Numerical (multiple fields) — pie chart; column charts
Active Dataviz community online, does do a lot of bashing — should be more on implementation, not so much on choice.
Always, keep in mind, what’s your message
- Excel 2016: New — Data Imports; New Chart Types — Box & Whiskers format available 🙂
- Excelcharts.com — Histogram.
- Book Recommendation: Kirk, Andy. Data Visualization: A Successful Design Process. Pack Publishing, 2012.
- Excel –> Select Data then click Recommended Charts (new 2016 feature, I think)
- Marcy Phelps handout on “choosing the right graphic”
Real time data & Gate Counter
Good definition of big data — more than you can handle in a single spreadsheet?
- Big Data — Circ Data — where do you pull meaning from?
- LibQual evaulations
- Quadrants visually displayed — attendee mentioned that. (Gartner quadrant?)
- Give data visual look — have someone with familiarity with data, and get feedback before submitting to stakeholders.
Goal: Crafting message. What data resonates with your audience.
- Stephanie Evergreen — Book Recommendation — Presenting Data Effectively.
- Data Viz Checklist from Stephanie Evergreen
Infographics vs Data Dashboard
Infogram — real time data (freemium)
Piktochart — reports & infographics; banners & presentations
If report is more engaging, people more willing to read, synthesize, read, understand.
- See at a glance.
- Telling story, choosing key pieces.
- Venngage.com another template driven infographic generator
- Map Viz — add-ins in excel
- Apps for Office.
- Excel has quick analysis option. Could be used with viz, but also analysis.
- Google.com Fusion Tables
- publiclibraries.com data
- Word clouds: Wordle; Tagul; Tagxedo
- Stemming option with Tagul
- Zotero Visualization tool? Zotero timeline
- Microsoft Power Bi
- Color choosing tools.
- Many, many more tools linked to on the workshop resources page.