Learning and The Art of Statistics

I think this might be the UK version of the cover design; much more preferable than the US version.

This summer I started reading David Spiegelhalter’s book, The Art of Statistics, but never got beyond the second chapter. I also had plans to learn d3 and do a personal visualization project. So much for plans. School ended and then I got bombarded with work.

Summer. Poof. Gone.

So now that I’m back at UM, I started reading The Art of Statistics, again…from the beginning. If anyone asked me about it, I would simply say that I don’t know much about it. For most of my twenty-some working years, my answer was accepted … until now.

Should, should, should…

Should designers learn to code?

Ever since I started designing websites the question, “Should designers learn to code” has been a never-ending debate. I used to think, Nah, leave it to the experts. Focus on your strengths—design! It was the easy answer; the one I, as well as many of my designer friends wanted to hear. Yet now, as a student in a STEAM program, I can say with certainty that a designer who can code has superpowers.

I’m trying to use a method of highlighting where yellow seems important and green are terms I need to learn and look up more if I don’t understand.

Should designers learn statistics?

I don’t think there is a “Pass Go” card for statistics if you want to practice data visualization ethically, truthfully, professionally. Of course you don’t need to be a statistician but as I’m learning, it sure helps you understand data. It helps you ask questions and more questions. I am convinced more than ever that as we move into this Fourth Industrial Revolution, citizens must have at least some amount of data literacy. Spiegelhalter says:

More data means that we need to be even more aware of what the evidence is actually worth.

David Spiegelhalter, Author, The Art of Statistics: Learning from Data

So yes, designers should learn statistics and The Art of Statistics is a great way to start. The format is wonderful because Spiegelhalter asks a question at the beginning of each “lesson”. He weaves jargon, concepts, history and process into stories from data which are stories about people and communities. What’s not to love? His writing isn’t dry. It is conversational, sometimes funny and always approachable. It is void of academic uppity.

Learning about distribution and logarithmic scale.

I decided to try a different approach to learning based on another book I’m also reading, Make it Stick. It’s a great book for teachers and students about learning. One takeaway was to break down anything you are learning into a series of steps or connections.

An excellent book for students and teachers.

That may seem obvious but apparently I went through life with with a different method for learning (memorization). With that in mind, I downloaded an upgrade to MindNode to try it out again (Thank you Qinyu for the reminder) to see if this process of creating a structure, a network, a series of relationships would help lodge some of these new terms and concepts into my long term memory. I’m also hoping this process makes learning statistics more enjoyable. So far, so good. (yay, me)

If you’re curious, you can click on the image or this link to see a better view of what I’ve started.

I’ve created mind maps on paper but what I like about MindNode so far is that I can move networks around easy peasy. The more I read, the more I process, I can go back and adjust as things become more clear.

Networks, Relationships, Connections

As a side note, I’m also learning how to design for artificial intelligence (or is it with?) and what has been interesting these past few weeks are how statistics, psychology, data visualization, artificial intelligence, and human-centered design are moving together and intersecting at various points as I process everything I am learning. The work I do (currently literature reviews) in the UX Lab continues to expose me to related terms and concepts.

As I move through The Art of Statistics, I’m getting early hints that I’ll be learning about analyzing performance for machine learning (regression models, algorithms, prediction). Cool. And recently, I learned about the theory of Connectivism.

There’s a path developing. Lightbulb moments.

We Lost Three Billion Birds

My husband and I have binoculars and a bird guide on a window ledge in our dining room. Sounds dorky, right? Well, I’m old enough to not care much about being a dork these days so onward.

We have a few bird feeders in our backyard and if I was home about this time of year and through the winter, we would be watching for birds and trying to identify them. Sometimes we would use our trusty guide book or use Cornell’s Merlin Bird ID app. There was also a time when we would send information about which birds showed up in our backyard.

So when I came upon this story in the New York Times about the loss of 3 billion birds since 1970 I was shocked, then deeply saddened, then alarmed.

Now, when I say shocked, I wasn’t surprised that there was a population loss. Climate change, pollution, deforestation, agricultural practices, invasive species … the list goes on. But really … danger, Will Robinson!

3 billion birds since 1970.

3 billion birds. 50 years. Gone.

Now, this graphic is beautiful. The colors, the structure, the information is clear and helpful. This is the online version in the Science Section:

Birds Are Vanishing From North America – The New York Times

Clearly, hit hardest are the Grasslands. The article mentions that even Robins and Sparrows, and Starlings (those invasive species) experienced significant declines.

Robins and sparrows.

a full-blown crisis.

David Yarnold, president and chief executive of the National Audubon Society

Isn’t everyone shocked and alarmed?

Can you imagine a world without birds? I don’t know about you but that gives me the willies. In fact I recall a moment while walking in our neighborhood when I realized there were no bird sounds. It was the eeriest feeling. I went out every morning listening specifically for a chirp or a call. What a relief when I finally heard one.

Conjuring a world without birds is a thing I don’t dare imagine, like the death of a child. Their fate is our own.

Joel Sartore Photographer

Can you picture 3 billion birds?

So, while writing this post, I realized that I wanted to show you what losing 3 billion birds looks like. Can you picture it? I can’t. If I had to explain to my 7-year-old niece what a loss of 3 billion birds looks like what would I show her?

I couldn’t find anything but this piece, Drowning in plastic: Visualising the world’s addiction to plastic bottles by Reuters came to mind.

It is insane to see the relationship of plastic bottles to iconic structures like the Eiffel Tower.

1.3 billion bottles. “Every day the equivalent of a bottle pile half the size of the Eiffel Tower in Paris is sold around the world.” Reuters

Would the story have greater impact and meaning among non-bird experts and fans if there was a way to see scale; a tasteful rendering and relationship of the number of birds to something familiar like the Statue of Liberty or covering an area the size of Texas or ?

I’m thinking yes. Do I have an exact solution? No, but I’d love to explore it some time in the future. In the meantime, I’m going to continue to help conservation efforts and take every day I see birds (even those pesky Starlings) as a gift.


Project Pitch: The Poetry of Anne Sexton

Today we pitch our project proposals.

Time limit: 3 minutes.

Writing poems, my quiet place.

I signed up for a poetry class during summer break while an undergrad at The University of the Arts (UARTs) pursuing my BFA in Photography. I had no idea what I was getting into but I went with it. The experience turned out to be one of the unexpected “bests” of my life. That summer was also the year I learned, in great detail what happened to my grandfather , a man I never knew because someone decided to take his life long before I was ever imagined.

Writing poems gave my anger and loss (can you mourn someone you never knew?) a place to rest. It was also the first time I recited a poem in public. Granted, it was a room of my classmates but for someone who honestly prefers to sit in the audience, this was big. Huge.

Transformations was my introduction to Anne Sexton.

It’s been awhile since I’ve read her poems and this idea of a project has been a wonderful way to reacquaint myself with her and her work. It’s been, well, decades. Does her work resonate with me now as it did then? I believe this project will be an emotional journey as much as it is a learning experience. What will I discover?

I’m proposing a visualization that slightly petrifies me.

I tweeted last night that I’m already in the “dark swamp of despair”.

Self-learning: Python, D3, scraping data, build a corpus… oh my.

Blogs, tweets, StackOverflow, GitHub, Codecademy, Lynda … you name it, I’m searching for “How to …” often.

  • Learning python: New jargon. Decisions about IDEs. Pandas… they aren’t cute black and white furry animals?
  • Learning how to properly analyze and clean text data: Pre-processing? Does my data need to be organized in a csv file or like a massive dump in a txt file? Looking for tutorials on how to prep text with Jupyter notebook) Soooo many questions.
  • Learning D3 is not going to be easy though I am hopeful Amelia Wattenberger’s book, Full Stack D3 and Data Visualization will be a huge help. (She is based in Rochester, NY – cool.)

Some classmates have suggested that I use R instead of Python. I’ve gone back and forth on this. Maybe I’m crazy but I prefer to learn Python because it seems a language that crosses many disciplines for many applications. Then, there’s the fact that python was named in honor of Monty Python’s Flying Circus!

Part of this is also recognizing that I know what is good but my abilities to execute are far below what it probably takes to meet my own expectations. Ira Glass has something to say about this and I’m trying to find some comfort. One would think familiarity of this mind space would make it easier each time. But no…

Still, I have some bright sides.

Mindy McAdams, a data journalism professor at the University of Florida was helpful in getting me setup with Python. I didn’t know it at the time but miniconda was the right way to go and given the number of IDE options, I’m so glad I went with Jupyter Notebook. (Seriously Mindy I cannot thank you enough.)

The bright side wouldn’t be complete without mad props to Lenny Martinez, one of UM Interactive Media’s data journalism professors. His positivity is all goodness when I feel like I’m drowning.

And, to end on a positive note, I do know how to install packages and I know kernals and cells, virtual environments and I’m not afraid of Terminal so much anymore. Not bad. Perhaps I need to just get over the fear that I’ll break something if I type incorrect syntax and also not worry so much about creating a properly formatted text file.

Ah, the “known knowns, known unknowns, and unknown unknowns”.

Thank you, Shirley Wu

Two mailing addresses, a few missed issues of Net Magazine.

Missing magazines show up—yay!—but left unopened for weeks.

Finally open the envelope, glance at the covers and they sit for another week. I was beginning to feel that read-your-New-Yorker-magazine-pressure every time I passed them.

But just yesterday when I sat down to write a blog post about a yield curve I saw (saving that for later), I noticed a small cover line on the August 2019 issue of Net Magazine.

Learn from a Data-Viz Whizz

Really? Cool.

Who is the Data-Viz Whizz?

Flip, flip, flipShirley Wu.

Opening spread in the Voices section of Net Mag, August 2019.

Shirley Wu, one half of the popular Data Sketches project, creates highly interactive, beautiful data visualizations. Here, she gives us a look behind the scenes and shares the lessons she’s learned.

Intrigued, I immediately sat down to read about Shirley Wu.

It was an inspiring read. I love stories about people who make changes; that she switched from being a front-end software engineer to become a data visualization designer is cupcake.

So, do you recall periods in your life when you are planning your next step or struggling or hoping for something but not sure what and the universe sends you a person or a moment or a sign to help you take that one small step forward?

This interview with Shirley is one of those moments. Noticing that small cover line near the UPC symbol is nearly impossible but I did and that led me to discovering Beautiful for the first time. Why is this important? Because I’m planning to tackle my first interactive data visualization of poems (Pablo Neruda? Anne Sexton? Maya Angelou? — I need to decide) and it helps to see what other data viz designers have done using text. Part of learning is seeing and understanding what is possible. It may take me some time to reach Shirley’s level of talent but her work and her words were a spark.

Correction: In my excitement, I had flipped Shirley Wu and Nadieh Bremmer’s work in my mind. My sincere apologies to both women. Nadieh is the designer behind Beautiful and Shirley is the designer of Explore Adventure (below), an equally beautiful (see what I did there?) and fun visualization about the travel search connections between countries, seasons, attractions, and more.

Screengrabs from Explore Adventure. This section about searches for Qin Shi Huang was the most interesting; however, I’m trying to still understand how I’m supposed to know the searches happen during spring, summer, winter or fall when just looking at the visualizations.

Lesson learned? Give yourself enough time to triple check your work, what you read, and own up to your mistakes.

Beautiful visualization interaction
The snaking animation is delightful and on hover, I learned more.

What I love about Beautiful is the overall simplicity, the subtle animations and the surprising level of detail and information which isn’t obvious at first. It’s fun and interesting. My only wish: a little more feedback during my interaction with the top 10 words per language.

Beautiful visualization
I wanted a slight animated fade or subtle change in value when I hovered over any of the shapes for each country.

Legends is just stunning and ok, I admit, I have a thing for that color palette. The cool factor is huge. Immediately I wished for Legends to be realized into a physical space that I could walk in and around (VR anyone?) with additional layers of information as I interact with each crystal.

Legends visualization from above
This feels like you are flying just above the surface of an alien planet (I watch a lot of Sci-Fi).
Legends visualization side view
I imagined myself walking among these crystals. What if I could touch them and they would light up and reveal something interesting?

I also admire the fact that she shares her knowledge about D3.js with others through workshops (with live coding!), user groups and online courses. She mentions her process and a lot of the tools she uses to clean, understand, explore, prototype, and design. It is a list I’m definitely planning to check out as I begin my journey learning D3.js through Coursera and making my first interactive visualizations.

So, thank you, Shirley Wu for making feel even more excited about data visualization, sharing lessons you’ve learned, your take on tools, and showing me what is possible with D3.js.

I cannot wait to see more of what you create.

PS: It’s always nice to read about people you know, especially when you are also learning from the same. She mentions Alberto twice. The first when he invited her to dive into data that would result in Beautiful and the second when she mentions how teaching forced her to learn so she read a few books. The Functional Art, she says, is “one of my favourites”. Cool.

Fonts for Data Visualizations

Professor Cairo has a voracious appetite for reading and he thankfully likes to share books and articles. One article he included, Finding the Best Free Fonts for Numbers was an interesting read as I have a thing for type and fonts. I get picky and can spend probably too much time selecting a font that I feel works well. I’ve also taught typography classes so while I am not a type expert, I am knowledgeable about typography.

In general, I agree with the list Samantha recommends. Not everyone can afford some of the best designed super families out in the wild. I also agree that free fonts aren’t always the best choice. Most are poorly designed and more importantly were probably created for the most generic of applications. So, again, she has compiled a thoughtful list.

Old Standard TT

I disagree about one typeface: Old Standard TT.

I do not think it would function well for data visualizations where type sizes are below possibly 14 points and that might be generous. Why? Old Standard TT can be quite interesting at large sizes; however, at smaller sizes, it starts to fall apart.

I need reading glasses to be able to read Old Standard TT at 14 pts. Even with it set in black on a white background (great contrast between figure and ground), it is quite challenging to read. Imagine if it is set in a color also on a colored background. Personally, if it is hard to read, I won’t. In my mind, that is the worst possible user experience.

My recommendation: If you want to use Old Standard TT, use it for display copy—headlines, subheads, or instances where you want to set a numeral in a particularly large size.

Oldstyle or Lining

Samantha’s recommendation for lining and tabular is a good base; however, this should not be a hard and fast rule. Why? Because there is a purpose for Oldstyle figures. Oldstyle figures work well when used with running text. They don’t interrupt the flow of reading because they share the same x-height as their lowercase character companions. Lining numbers in contrast stand out when sharing the same baseline as lowercase characters in running text.

Oldstyle figures can be used for data visualizations especially in places where numbers share the same baseline as text. For example, annotations. They are also readable in tables and other data visualizations purposes. Oldstyle figures can also be tabular so please, don’t rule out a typeface because they have oldstyle figures.

OpenType and Investing in Typefaces

With OpenType fonts, you get the best of all worlds, usually. For figures, OpenType give you the flexibility of setting figures in tabular and lining and tabular and oldstyle. Usually a designer can also set type as proportional as well. This is one of the perks of OpenType fonts and investing in building a library of high-quality typefaces. (Use a font manager such as Extensis’s Suitcase Fusion). Many free fonts are not OpenType.

My Favorite Fonts for Data Visualizations So Far…

Below is a short list of sans serif typefaces I use over and over again. Many are large families so you also have a choice of many styles: thin, light, italic, regular, bold, etc.

Benton Sans

Fira Sans




Poynter Gothic Text

PT Sans

Nimbus Sans


News Gothic

FF Meta


If you have an Adobe CC subscription …

Many of the fonts above are available through fonts.adobe.com (formerly TypeKit). It’s one of the perks of having an Adobe Creative Cloud subscription. If you are interested in others, try a search for sans serif with a large x-height. (A larger x-height usually means greater readability at smaller sizes.)

If you want to learn more about typography, I highly recommend Ellen Lupton’s website and book, Thinking with Type. I also have plenty of books which I’ll try to share soon. The great part about owning high-quality typefaces: you don’t need many. This is what makes OpenType super families so appealing.