Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca - contact me Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca on Twitter Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca - Lumondo Photography Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca - Pi Art Martin Krzywinski / Genome Sciences Center / mkweb.bcgsc.ca - Hilbertonians - Creatures on the Hilbert Curve
And whatever I do will become forever what I've done.Wislawa Szymborskadon't rehearsemore quotes

luminance: fun


In Silico Flurries: Computing a world of snow. Scientific American. 23 December 2017


visualization + design

Brewer Palettes

Brewer Palettes at a Glance

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
All the Brewer palettes: qualitative, sequential and diverging. For each palette (e.g. spectral) the source colors are shown as well as all its n-color subsets. (zoom)

Presentation About Color and Brewer Palettes

If you're new to Brewer palettes, or color, catch up with this presentation. Color palettes matter - Brewer palettes and perceptual uniformity - Martin Krzywinski

COLOR NAME DATABASE

I maintain a comprehensive database of named colors (3,116 colors), compiled from a variety of color name lists.

Visualization and Perception

Why Should Engineers and Scientists Be Worried About Color? by Bernice E. Rogowitz and Lloyd A. Treinish (IBM Thomas J. Watson Research Center, Yorktown Heights, NY).

Perception in Visualization by Christopher G. Healey (Department of Computer Science, North Carolina State University)

LAB and LCH gradient picker

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
Interactively create LAB and LCH color gradients interpolated across any number of colors.

Lch and Lab colour and gradient picker is a great tool by David Johnstone. It's a great way to generate color ramps—go ahead, go crazy!—and compare how the ramps look in different color spaces. Shame on you, HSV!

PaletteView — create continuous Brewer palettes

PaletteView is an exceptional tool by Magnaview to create continuous Brewer palettes. This tool is described in [1] and operationalizes Cyntha Brewer's color selection method into an algorithm that selects customizable color palettes from LCH space.

[1] Wijffelaars M, Vliegen R, Van Wijk JJ et al. 2008 Generating Color Palettes using Intuitive Parameters Computer Graphics Forum 27:743-750.

Brewer Palette Adobe Swatch Files

You can import Brewer palettes into Adobe applications such as Illustrator, Photoshop and InDesign using either the .ase or .ai swatch files.

download

Brewer palette ase swatch file for Adobe Illustrator Brewer palette ai swatch file for Adobe Illustrator Brewer palette pdf color file Brewer palette txt color file

install

In Illustrator, load the swatches from the swatch window menu. The swatch window can be accessed using Window > Swatches.

Select Open swatch library

then choose Other library...

and load either the .ase or .ai file — both contain the same content.

Brewer palettes are color combinations selected for their special properties for use in data visualization and information design.

The challenge

Selecting effective colors for bar plots, pie charts, and heat maps is made more difficult by the fact that the way we select color in software does not reflect how we perceive the color.

There are many examples of poor color combinations in published figures. For example, if categories are encoded with a combination of bright and dark colors, the bright colors will dominate the reader's attention. On the other hand, if two colors appear similar, the reader will instinctively perceive them as belonging to a group and infer that the underlying variables are related.

Colors with poor contrast (colors with similar perceived brightness) or simultaneous contrast (pure colors) also interfere with interpreting figures.

Selecting Colors in RGB and HSV

Most people select colors using RGB sliders, which is just about the worst way to pick a color! Consider the fact that when we look at a color, we cannot easily decompose it into its red, green and blue components. This limits usefulness of RGB for color selection.

HSV is a better color space, which defines a color based on hue, saturation and value. These are three properties that we intuitively assess when we see a color. We think of a "dark rich blue" and "light faded red", making HSV a reasonably useful model for color selection. Unfortunately, HSV has a nagging problem — although it is based on intuitive parameters, it is not perceptually uniform.

Perceptual Uniformity

A color space that is perceptually uniform defines colors based on how we perceive them. Distances between colors in the space are proportional to their perceived difference.

Above, we saw that HSV was not perceptually uniform. Moving the hue slider by 60 can have a small or large effect on a color, depending on where the slider is positioned.

Consider the following example. You have a chart that uses two colors, and orange and green. Both were chosen with S=V=100%. You now need to select a second color for each that is brighter. You cannot directly use HSV because both orange and green colors are already at full value. How do you intuitively increase brightness?

The reason why you cannot in do this in HSV is because V does not directly correspond to the color's perceived brightness. You are stuck fiddling with the saturation and value to try to select a brighter pairing.

What would be useful here is a color space which uses the intuitive parameters of HSV, but is perceptually based. In other words, instead of value, the space would define a color based on its perceived brightness. Luckily, this space exists — LCH, which defines color based on its luminance (perceived brightness), chroma (purity) and hue. Unfortunately, design and presentation software do not have LCH sliders and we cannot easily take advantage of this color space.

This is where the Brewer palettes come in.

Brewer Palettes

Brewer palettes were selected for their perceptual properties. These palettes were created by Cynthia Brewer for the purpose in cartography, but have found use in other fields.

Types of Brewer Palettes

There are three types of Brewer palettes

  • qualitative — colors do not have a perceived order
  • sequential — colors have a perceived order and perceived difference between successive colors is uniform
  • diverging — two back-to-back sequential palettes starting from a common color

Swatches of Brewer Palettes

I have prepared Brewer palette swatches in .ase or .ai format. For programming, use the plain-text version.

The image below (zoom) shows all the Brewer palettes.

Brewer palette colors - all swatches

Uses of Brewer Palettes

Qualitative palettes are excellent for bar plots and pie charts, where colors correspond to categories.

Grayscale Brewer palettes are available and are perfect for achieving good tone separation in black-and-white figures.

Sequential and diverging palettes are useful for heatmaps.

Brewer Palettes and Color Blindness

Some Brewer palettes are safe for color blindness — the pink-yellow-green (piyg) is one. For others, see colorbrewer.

I have designed 15-color palettes for color blindess for each of the three common types of color blindness.

VIEW ALL

news + thoughts

Statistics vs Machine Learning

Tue 03-04-2018
We conclude our series on Machine Learning with a comparison of two approaches: classical statistical inference and machine learning. The boundary between them is subject to debate, but important generalizations can be made.

Inference creates a mathematical model of the datageneration process to formalize understanding or test a hypothesis about how the system behaves. Prediction aims at forecasting unobserved outcomes or future behavior. Typically we want to do both and know how biological processes work and what will happen next. Inference and ML are complementary in pointing us to biologically meaningful conclusions.

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
Nature Methods Points of Significance column: Statistics vs machine learning. (read)

Statistics asks us to choose a model that incorporates our knowledge of the system, and ML requires us to choose a predictive algorithm by relying on its empirical capabilities. Justification for an inference model typically rests on whether we feel it adequately captures the essence of the system. The choice of pattern-learning algorithms often depends on measures of past performance in similar scenarios.

Bzdok, D., Krzywinski, M. & Altman, N. (2018) Points of Significance: Statistics vs machine learning. Nature Methods 15:233–234.

Background reading

Bzdok, D., Krzywinski, M. & Altman, N. (2017) Points of Significance: Machine learning: a primer. Nature Methods 14:1119–1120.

Bzdok, D., Krzywinski, M. & Altman, N. (2017) Points of Significance: Machine learning: supervised methods. Nature Methods 15:5–6.

...more about the Points of Significance column

Happy 2018 `\pi` Day—Boonies, burbs and boutiques of `\pi`

Wed 14-03-2018

Celebrate `\pi` Day (March 14th) and go to brand new places. Together with Jake Lever, this year we shrink the world and play with road maps.

Streets are seamlessly streets from across the world. Finally, a halva shop on the same block!

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
A great 10 km run loop between Istanbul, Copenhagen, San Francisco and Dublin. Stop off for halva, smørrebrød, espresso and a Guinness on the way. (details)

Intriguing and personal patterns of urban development for each city appear in the Boonies, Burbs and Boutiques series.

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
In the Boonies, Burbs and Boutiques of `\pi` we draw progressively denser patches using the digit sequence 159 to inform density. (details)

No color—just lines. Lines from Marrakesh, Prague, Istanbul, Nice and other destinations for the mind and the heart.

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
Roads from cities rearranged according to the digits of `\pi`. (details)

The art is featured in the Pi City on the Scientific American SA Visual blog.

Check out art from previous years: 2013 `\pi` Day and 2014 `\pi` Day, 2015 `\pi` Day, 2016 `\pi` Day and 2017 `\pi` Day.

Machine learning: supervised methods (SVM & kNN)

Thu 18-01-2018
Supervised learning algorithms extract general principles from observed examples guided by a specific prediction objective.

We examine two very common supervised machine learning methods: linear support vector machines (SVM) and k-nearest neighbors (kNN).

SVM is often less computationally demanding than kNN and is easier to interpret, but it can identify only a limited set of patterns. On the other hand, kNN can find very complex patterns, but its output is more challenging to interpret.

Martin Krzywinski @MKrzywinski mkweb.bcgsc.ca
Nature Methods Points of Significance column: Machine learning: supervised methods (SVM & kNN). (read)

We illustrate SVM using a data set in which points fall into two categories, which are separated in SVM by a straight line "margin". SVM can be tuned using a parameter that influences the width and location of the margin, permitting points to fall within the margin or on the wrong side of the margin. We then show how kNN relaxes explicit boundary definitions, such as the straight line in SVM, and how kNN too can be tuned to create more robust classification.

Bzdok, D., Krzywinski, M. & Altman, N. (2018) Points of Significance: Machine learning: a primer. Nature Methods 15:5–6.

Background reading

Bzdok, D., Krzywinski, M. & Altman, N. (2017) Points of Significance: Machine learning: a primer. Nature Methods 14:1119–1120.

...more about the Points of Significance column

Human Versus Machine

Tue 16-01-2018
Balancing subjective design with objective optimization.

In a Nature graphics blog article, I present my process behind designing the stark black-and-white Nature 10 cover.

Nature 10, 18 December 2017