7
$\begingroup$

I'm looking for some references on creating effective graphs/data visualizations.

I've found a bunch of books that show how to create data visualizations using certain tools (like R/ggplot vs python/pandas) but that's not really what I'm looking for. I'm looking for a reference that explains different types of charts with respect to stats/math. I want more theory than process.

I want to know the different types of charts and how to use them. Anything helps!

$\endgroup$
11
$\begingroup$

I think that the work of William Cleveland is going to be closer to what you want that that of Tufte. Cleveland wrote two books:

  1. Visualizing Data (1993)
  2. The Elements of Graphing Data (1985)

The first book, in particular, may be what you want. Here is a publisher's description:

Visualizing Data is about visualization tools that provide deep insight into the structure of data. There are graphical tools such as coplots, multiway dot plots, and the equal count algorithm. There are fitting tools such as loess and bisquare that fit equations, nonparametric curves, and nonparametric surfaces to data. But the book is much more than just a compendium of useful tools. It conveys a strategy for data analysis that stresses the use of visualization to thoroughly study the structure of data and to check the validity of statistical models fitted to data. The result of the tools and the strategy is a vast increase in what you can learn from your data. The book demonstrates this by reanalyzing many data sets from the scientific literature, revealing missed effects and inappropriate models fitted to data.

An even more theoretical book is The Grammar of Graphics by Leland Wilkinson. The description:

This book was written for statisticians, computer scientists, geographers, researchers, and others interested in visualizing data. It presents a unique foundation for producing almost every quantitative graphic found in scientific journals, newspapers, statistical packages, and data visualization systems. While the tangible results of this work have been several visualization software libraries, this book focuses on the deep structures involved in producing quantitative graphics from data. What are the rules that underlie the production of pie charts, bar charts, scatterplots, function plots, maps, mosaics, and radar charts? Those less interested in the theoretical and mathematical foundations can still get a sense of the richness and structure of the system by examining the numerous and often unique color graphics it can produce. The second edition is almost twice the size of the original, with six new chapters and substantial revision. Much of the added material makes this book suitable for survey courses in visualization and statistical graphics.

This book is very theoretical.

$\endgroup$
  • 2
    $\begingroup$ 2nd edition of the Elements book 1994. I echo this endorsement of Cleveland strongly. Tufte is great but Cleveland speaks more directly and in much technical detail to anyone statistically minded. I will add that these books really do not date in any fundamental sense. $\endgroup$ – Nick Cox Feb 14 '17 at 13:05
  • 2
    $\begingroup$ There is an over-arching (or under-pinning) theory in Wilkinson's book, which is best commended by the fact that Hadley Wickham built on it in designing his ggplot2 in R. But it's also a great book to skip and skim. $\endgroup$ – Nick Cox Feb 14 '17 at 13:07
6
$\begingroup$

Look at the series of books written by Ed Tufte. They are discussed by wikipedia in the article https://en.wikipedia.org/wiki/Edward_Tufte.

  1. The Visual Display of Quantitative Information. 1983; Second Edition 2001. Cheshire, CT: Graphics Press

  2. Envisioning information. 1990. Cheshire, CT: Graphics Press

  3. Visual Explanations: Images and Quantities, Evidence and Narrative Graphics Press. 1997. Cheshire, CT: Graphics Press

  4. Beautiful Evidence. 2006. Cheshire, CT: Graphics Press

$\endgroup$

We're looking for long answers that provide some explanation and context. Don't just give a one-line answer; explain why your answer is right, ideally with citations. Answers that don't include explanations may be removed.

  • $\begingroup$ I did give a reference and my answer is not too short. $\endgroup$ – Michael Chernick Feb 13 '17 at 23:13
  • 1
    $\begingroup$ We expect answers to "list-of" questions like this one to include, at a minimum, a cogent reason for the recommendation. Answers that do not supply reasons are usually deleted or converted to comments. $\endgroup$ – whuber Feb 13 '17 at 23:18
  • 1
    $\begingroup$ @whuber I gave a very appropriate answer referencing the three books written by Edward Tufte. Do you have a better suggestion? $\endgroup$ – Michael Chernick Feb 13 '17 at 23:21
  • 1
    $\begingroup$ Yes. First indicate each book separately, by title. Along with those indications describe how that book recommendation responds to the question. What theory or theories does Tufte advance? From what particular perspective? Why would it be worthwhile to consult these texts? How do they differ among themselves? Etc., etc. $\endgroup$ – whuber Feb 13 '17 at 23:23
  • 2
    $\begingroup$ I've added the second of four (to date) of Tufte's self-published books (note that graphics is a secondary theme in his earlier books). I won't try to impute Michael's commendation. $\endgroup$ – Nick Cox Feb 14 '17 at 13:09
5
$\begingroup$

At the risk of being crucified, I would advise against Tufte, Wilkinson, Cleveland etc. and all other classics if you're just starting out.

The reason is the following objective laid out by you (emphasis added):

I'm looking for some references on creating effective graphs/data visualizations.

So even though you don't explicitly want language dependent books/tutorials, you want your knowledge to be applied rather than an abstract theoretical exercise over coffee. Starting with what I call the classics is like reading Shakespeare because you want your language to be more eloquent. The discussions in the books are excellent for laying the foundations to understand effective data visualization; but considering the technological advancements up to today - the books aren't much help in developing the applied bent of mind (Grammar of Graphics- Wilkinson being the slight exception because of the relevance to ggplot2 but in that case I would advise reading works of Hadley Wickham, the package author instead).

Some good resources you could look at are FlowingData (Nathan Yau), Perceptual Edge (Stephen Few) and Storytelling with Data (Cole Knaflic) and the books by the blog authors. The reason being as follows:

  1. These works already encompass the research from the classics
  2. The language is less academic and easier to understanding
  3. The regularly updated blogs act as supplemental material to the books

It's a pity Aaron Koblin hasn't published any books about his unique take on large data visualizations.

I do not discount how useful Tufte, Cleveland and Wilkinson's work is, but after toiling through a few of them and still only being marginally better at modern data visualization tools, Stephen Few's "Show me the Numbers" was like a light switch went on.

$\endgroup$
4
$\begingroup$

It depends strongly on the language you prefer. As I am not using Python for data visualisation frequently I can only recommend you books relating to data visualisation in R. After writing this post I reread your question and Nr. 1, Nr. 2 and maybe Nr. 4 might be the most theoretical. Though Nr. 6 also explaines you theoretical aspects it is specialised on visualising unsupervised machine learning techniques.

  1. R Graphics by Paul Murrell

The author Paul Murrell has a significant part in developing the graphics of the R language. He developed the "Grammar of Graphics" concept which is the concept underlying the ggplot2 library. The book is rather advanced although you do not need a lot of preknowledge necesarrily and pretty theoretical. It is the best book for people who genuinely want to understand the concepts of data visualisation in R, but I do not recommend it for beginners.

  1. HTML Widgets

Is a must for interactive data visualisation. Various JavaScript libraries are translated into and adapted to R. You can include most Widgets in RShiny, Markdown (rendered as HTML) or in the console). My favorite HTML Widgets are

  • Plotly (A library on interactive data visualisation which is also available for various other languages such as Python and Matlab)
  • Leaflet (interactive visualisations with Maps)
  • dygraph (which offers a broad variety for interactive time series visualisation)
  • datatable (written by Yuhui Xe from RStudio who also wrote the knitR and the bookdown package. Prolific for showing tables))

    1. Guide to create beautiful graphics in R

This book is rather beginner friendly. Its examples are primarily shown in ggplot2. When I started learning advanced data visualisation techniques in R I primarily used this one and the official ggplot2 website.

  1. The official ggplot2 website

Is the best starting point to learn ggplot2, but it can appear overwhelmingly if you are not willing to be passionate and if you don't have a lot of time. ggplot2 is awesome, but it can have a steep learning curve, e.g. you cannot write the "+" at the beginning of the line. All theoretical concepts are also explained.

  1. Official Shiny gallery

Shiny is the most used R-library for building up apps with R. It can be substituted by BI tools like Tableau or Qlickview. shinyjs is a great extension of shiny which combines shiny with javascript, but you can also include HTML, CSS and JavaScript on your own.

  1. Cluster Analysis in R

This book comes from the same authors as the Guide to beautiful graphics (nr.3). It is a specialized book for visualising unsupervised machine learning techniques and particularly clustering.

7.Easy tutorial

In case you just start visualising and I overwhelmed you a little bit.

$\endgroup$
0
$\begingroup$
  1. R for Data Science by Garret Grolemund and Hadley Wickham

  2. Top 50 ggplot2 Visualizations

  3. The R Graph Gallery

  4. r4stats.com

$\endgroup$

We're looking for long answers that provide some explanation and context. Don't just give a one-line answer; explain why your answer is right, ideally with citations. Answers that don't include explanations may be removed.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.