Ask HN: What is the best book on data visualization in 2021?

[zwaps]

I am looking for a comprehensive resource on data visualization. There are many individual resources that have been posted over the years (such as in blog posts). However, as a non-expert I am not sure which advice is good to follow (e.g., many blogs are published by companies trying to sell me a particular product or package).

Not sure if the book is the ideal format for this, but it seems to have the highest chance of being competent, unbiased and comprehensive. For the latter point, I think important topics are

- Colors and accessibility

- Concrete examples of color schemes

- Pros and Cons of different visualizations vis-à-vis data

- Print versus digital

- Fonts and layout considerations

- Perhaps short insights into the research on these points

But perhaps there are other things not on my radar that are current or interesting?

[nothrowaways]

I did a quick research on this topic sometime ago and the following was my conclusion.

1. The Visual Display of Quantitative Information - Edward Tufte

2. Storytelling With Data

3. Documentation pages of data Viz tools e.g. 1, 2, 3 depending on your programming language of choice.

IMHO python ecosystem is still struggling with solid visualization as compared to JavaScript and R. Ggplot gives you shiny Viz but I'm not a fan of R.

D3 and it's derivatives are awesome for dynamic Viz if that's what you are after. GL

1. https://github.com/d3/d3/wiki

2. https://ggplot2.tidyverse.org/reference/ggplot.html

3. https://matplotlib.org/3.5.0/index.html

[Foivos → nothrowaways]

For python this open access book is excellent: https://github.com/rougier/scientific-visualization-book

[jszymborski → Foivos]

I can second this! Really no better reference for creating plots in python.

[ekianjo → nothrowaways]

> 1. The Visual Display of Quantitative Information - Edward Tufte

Good book but it starts to show its age. We need a newer version regarding more recently developed visualization techniques.

[kuhewa → ekianjo]

He just released a fifth book. Don't know how it rates. None of his were particularly about the technical side of production though

[hx2a → kuhewa]

For anyone else looking for the 5th book:

https://www.edwardtufte.com/tufte/seeing-with-fresh-eyes

[zcw100]

William Cleveland, "Elements of Graphing Data"[1] and "Visualizing Data"[2]

Cleveland worked at Bell Labs and his work is timeless and very practical. It isn't about artwork or making things look nice. It's about communicating information. I like to contrast his work with Tufte's. Tufte comes up with some nice stuff but I find it to be more art than science where the information is clearly sacrificed to ascetics. With visualizations I always ask myself, "Do you know more now about what is going on than you did before or is it just something nice to look at?" Way too often it's the latter.

[1] https://www.amazon.com/Elements-Graphing-Data-William-Clevel...

[2] https://www.amazon.com/Visualizing-Data-William-S-Cleveland/...

[jstx1]

The thing that's specific to data visualisation is the problem of mapping properties of your data to properties of the visualisation. Some good resources on this are

The Visual Display of Quantitative Information - Edward Tufte

The Grammar of Graphics - Leland Wilkinson

FlowingData's guides - https://flowingdata.com/category/guides/

But from your list I gather that you're more interested in the visual design of things (colour schemes, fonts etc); if that's the case you can search for resources on visual/graphic design which I don't know too much about. I can say that the choice of fonts and colours is a small part of data visualisation - very few basic princpiples, using a tool with good defaults and some common sense get you a very long way. The bigger part is deciding which parts of your dataset to communicate in what way and the resources that I mentioned are pretty good at teaching you that.

[brezelnbitte]

Tufte is a waste of time unless you’re interested in history of dataviz. While he makes excellent points on the graphics of the time, he makes at least an equal portion of questionable points. He's a bit of a luddite when discussing the utility of computers and takes a dogmatic approach to minimalism and design that doesn't have room for differing opinions. His opinions also do not hold in light of recent research into how humans interpret visual information. For example, he often recommends tables over bar charts when it's now been shown that bar charts are processed faster and easier.

Better Data Visualizations by Jonathan Schwabish is the best. He combines all the insight from recent data biz books like Storytelling with Data or Cairos series into a comprehensive guide. Even has a practical section on visualizing qualitative data, which I never see.

[fractallyte]

Nadieh Bremer's blog: https://www.visualcinnamon.com/

And she co-wrote a beautiful book with Shirley Wu: https://www.datasketch.es/

[vladsanchez → fractallyte]

Never heard of them. Beautiful and artful. Thanks for sharing.

[temp234 → vladsanchez]

If you enjoyed their work also check out the work of http://giorgialupi.com/ and https://www.accurat.it/

[apwheele]

While I really love Tufte's books, for more direct advice I recommend Albert Cairo's The Functional Art. I don't remember discussion of fonts offhand, but everything else in your list is covered.

Several of the other comments, such as Wilkinson's GoG or Cleveland's research, I don't think make sense as an intro to the topic. I have posted my notes on various data viz books (which are quite heterogenous) in this blog post: https://andrewpwheeler.com/2020/11/04/overview-of-dataviz-bo...

[squid_demon]

(Slightly off-topic, as the link is a collection of research papers rather than a book.) A Visual Survey of Text Visualization Techniques: https://textvis.lnu.se/

[esturcke]

Colorgorical is neat (http://vrl.cs.brown.edu/color). It combines previous work on judging categorical color palettes based on discriminability/aesthetics/etc and presets a generator to find new palettes with various constraints.

[abby_3017]

This book is really details and explain minute things in this domain. Tamara Munzner. "Visualization Analysis and Design" https://www.cs.ubc.ca/~tmm/vadbook/

You should take a look into this.

[esturcke → abby_3017]

This is my pick as well. Very nice survey with an academic/formal approach. I enjoy the emphasis on understanding the what (data)/why (tasks) before getting to the how (vis). She also emphasizes analysis and validation.

She also had a workshop last IEEEVis that’s a nice summary of her book: https://m.youtube.com/watch?v=_sFzBmQeCLY

[kuhewa]

http://www.janvwhite.org/

[fractal618]

The Microsoft excel help file

[thunderbong → fractal618]

In my opinion, this is truly the best suggestion. Mirosoft Help is just fantastic. In Excel, I learnt VBA entirely using Microsoft Help.

Visualizations, walk-throughs, and in most cases, there will be a link to the exact setting to get something done.

The amount of effort spent in the documentation has to be really phenomenal!

[brylie]

Now You See It: An Introduction to Visual Data Sensemaking, Second Edition. Stephen Few, $49.00 MSRP, April 2021

http://www.stephen-few.com/nysi.php

[davidivadavid]

Can't go wrong with Tufte, but I'm also partial to Stephen Few for a more pragmatic approach ("Show Me the Numbers", "Information Dashboard Design").

[jstx1 → davidivadavid]

I was thinking of Few when writing my response but I couldn't recall his name. You're right, he's pretty good - http://www.perceptualedge.com/library.php#Articles

[ansupter]

To throw another hat into the ring, I liked "Designing Visual Interfaces" by Mullet and Sano. It has one of my favorite quotes:

> Without minimizing the value of intuition as a problem solving tool, we propose that systematic design programs are more valuable from a communication standpoint than ad hoc solutions; that intention is preferable to accident; that principled rationale provides a compelling basis for design decisions than personal creative impulse.

[nojito]

Nothing really holds a candle to storytelling with data.

Cole is an excellent teacher and is able to clearly share exactly what practitioners need to know to make meaningful visualizations.

[amrrs → nojito]

I've not read the book but I attended their workshop. I took some practical advice about data viz and started making serious changes in my Visualizations. It's helped. My coworkers even noticed the change and mentioned it.

[haraball]

I've enjoyed Data Visualization Handbook by Koponen and Hildén - https://datavizhandbook.info/

Take a look at the table of contents on that link, it could be a good fit for what you're looking for.

[jboynyc]

Kieran Healy's Data Visualization: A Practical Introduction https://kieranhealy.org/publications/dataviz/

[dwt204]

Definitely "The Visual Display of Quantitative Information" by Edward Tufte

[throwaway4good]

I really liked "Information Visualization - using vision to think" by Card, Mackinlay, and Shneiderman. I don't know if they ever updated it or there is something similar out there that is not 2 decades old.

[IvanaSays]

In addition to a lot of the recos here, Gene Zelazny's "Say it with charts" is an old school book that a manager of mine had us read in concert with Tufte. It's a good simple primer on charts.

[belfalas]

You definitely must get a copy of Tufte's books, like others have said.

The "Book of Circles" and "Book of Trees" are also good.

[high_byte]

not a book but a collection + framework: ObservableHQ based on D3