When I started publishing on Tableau Public earlier this year, I was looking for a unique way to brand my work. I went down the path of trying to create a fun blog name along with a custom logo and hated every iteration. I ultimately decided to go for simple and personal - I would use my handwritten name on my work.Read More
It happens every week - after I post a new visualization on Tableau Public and/or share the link via Twitter, I realize I forgot to do something. Whether it's a missing URL, forgetting the device designer, or default tooltip formatting, I seem to forget something.
Coincidentally, I finished reading the Checklist Manifesto by Atul Gawande (http://a.co/0RLtaEt) and thought I could apply some of the techniques to improve the quality of my work. As such, I created a discipline checklist - a list of things to consider before every post. My intention is for this to be a living document and hope to include your feedback/best practices as well.
Finally, special thanks to @acrahen, @DrexelPooja, @josh_tapley, @taawwmm, @prettylawful and @drawwithdata for your input in helping get this version published.
Data Visualization Checklist
☐ Did you include the source reference?
☐ Do you want to include personal branding?
☐ If yes and not using images, add a transparent image over the text to link to the url
☐ Do you want to make a mobile version?
☐ Did you format all Tooltips?
☐ Did you turn off the command controls?
☐ Did you want to block any clicks on the dashboard? (use a blank sheet)
☐ If the font is important, did you use images rather than text?
☐ Did you make sure your dashboard actions are working correctly and only impacting the desired sheets
☐ Do you want a stand-alone visualization? (add this after the URL: &:showVizHome=no)
☐ Did you hide your sheets (and delete any unused sheets)
☐ Did you make sure your fonts match across your dashboard
☐ Did you spell check your text/titles?
☐ Did you cover up the Open Street logos (sorry Open Street)
☐ Did you turn off map panning and zooming
☐ Do you want to post your dashboard directly or a blog post?
☐ Did you use a URL Shortener
Google = https://goo.gl/
☐ Did you mention @tableaupublic
☐ Did you make an image of the viz and attach to your tweet?
☐ Do you want to include a gif?
☐ Did you want to time your tweet for a specific day/time? (use tweetdeck.twitter.com)
☐ Did you use hashtag #MakeoverMonday
☐ Did you mention both @VizWizBI and @TriMyData
☐ Did you include a static image (not a gif!)
Pie charts are everywhere and the casual observer might not notice they are generally a poor way of expressing data. Ask anyone who works with data visualization regularly and they will have an opinion, often a negative one, on the pie chart. As you delve into the theory of data visualization, you will most likely experience many of the five stages of the Kübler-Ross model of death. Hopefully, you'll come out in the last stage, acceptance, rather than depression. If you are looking for specific reasons pie charts don't work well, here are a few fantastic resources (here, here, here and here). Although the title implies extreme measures, pie charts (and their similarly delicious cousin, the donut chart) do have some value use in data visualization, although very limited (see chart below).
Stage 1: Denial
No pie charts aren't that bad. Here is a really good one I found here. They work really well in showing parts to whole relationships.
Stage 2: Anger
If pie charts don't do well to display large groups or really close percentages, then why use them at all? Why are there so many labels, markers, colors and legends needed?
Stage 3: Bargining
Ok, I'll use a pie chart only if I have 3 or less groups and they are really obvious. It is ok to use them only when talking about mutual fund constructions. How about I use a donut chart instead?
Stage 4: Depression
If pie charts suck so bad, why does anyone use them?
Stage 5: Acceptance
Ok, I totally understand the shortcomings of pie charts and will only use them when appropriate. I understand that this is very rare, but something I can use in my data visualization toolbox.
It’s been a decade since Tool’s last album, 10,000 Days. While there have been rumors of a new album in the works for the past three years, I hope this is the year! While we wait, I performed an analysis of their four studio albums to get some insights on how the band has evolved lyrically and the album contents.
The trends suggest we’ll see about 10 or 11 tracks with songs getting slightly longer and at least two filler tracks (i.e. tracks with no lyrics, sound effects or a song intro). Lyrically, we can export more complex songs in multiple categories (syllables, word length, unique words, complex words, reading level, lexical density). Finally, from the sentiment analysis, the lyrics are trending to a possibly neutral or positive vibe. I guess this can de a demonstration that teenage angst does regress to the mean after a quarter of a century.