Adding two new styles to mplstyles #3190

CamDavidsonPilon · 2014-07-05T18:30:26Z

The first, 538.mplstyle is a replication of the plots used on FiveThirtyEight.com, Nate Silver's blog. I wrote an blog article on the replication here (with some samples).
The second, bmh.mplstyle, is a what is used in Bayesian Methods for Hackers. I think it also won the PlotOrNot contest from last year ¯_(ツ)_/¯?

cc @ChrisBeaumont

tacaswell · 2014-07-05T18:35:32Z

Can you squash these down to two commits?

CamDavidsonPilon · 2014-07-05T19:01:28Z

@tacaswell done! (and learned some new git too)

CamDavidsonPilon · 2014-07-05T19:22:47Z

lib/matplotlib/style/stylelib/bmh.mplstyle

+axes.color_cycle: 348ABD, A60628, 7A68A6, 467821, D55E00, CC79A7, 56B4E9, 009E73, F0E442, 0072B2
+
+legend.fancybox: True
+figure.figsize : 11, 8figure.dpi : 100


Error here, I'll fix

Update 538.mplstyle Update bmh.mplstyle fixing line in bmh.mplstyle

tacaswell · 2014-07-06T00:22:07Z

This is the first round of style sheets we have gotten so there is no precedent, but I see no reason not to merge them for 1.4.0. @tonysyu you have comments?

tonysyu · 2014-07-06T02:57:04Z

Thanks for pinging me @tacaswell. The styles themselves are quite nice, and it'd be great to have more stylesheets included. I'm not crazy about the naming: 538 would probably be better as 'fivethirtyeight' (to match all their branding) and 'bmh' is a bit too terse for me, but I don't have a better suggestion.

Actually, one constructive criticism: I don't think most public stylesheets should have DPI specified. We should prefer the user's personal DPI preference, and possibly their figure size preference as well. The exception would be for specific publication-specific stylesheets, which may have exacting specifications.

CamDavidsonPilon · 2014-07-06T14:24:17Z

@tonysyu, I've addressed those issues: removed dpi/figsize settings, and renamed to fivethirtyeight

Tillsten · 2014-07-06T17:43:56Z

Adding a section with the available styles to https://github.com/matplotlib/matplotlib/blob/master/doc/users/style_sheets.rst would be wonderful. Even better with a picture per style.

CamDavidsonPilon · 2014-07-06T19:03:47Z

mmm, would that be better on the matplotlib website? I see those docs purely for creating new styles, not seeing what's available.

tacaswell · 2014-07-06T19:09:18Z

Those docs are what are used to generate the website.

CamDavidsonPilon · 2014-07-06T20:05:17Z

Best place to put the images in /doc/_static? Thoughts on the size of the images?

tonysyu · 2014-07-06T20:30:59Z

Actually, I think a plot script, rather than an image, would be preferable. Currently, there are style examples in the gallery here:
https://github.com/matplotlib/matplotlib/tree/master/examples/style_sheets

tonysyu · 2014-07-06T20:34:18Z

I should note that the plot should be fairly simple; something closer to plot_dark_background.py in that section, rather than the other two. Long term, it might be nice to have style sheets run with some "standard" plot for easy comparison (and less code).

tacaswell · 2014-07-06T20:38:05Z

Ideally we would have one chunk of code that would just loop over all of the style sheets an generate a figure. That requires understanding a bit more of the sphinx build system than I currently do.

tonysyu · 2014-07-06T20:52:53Z

Sorry, I wasn't very clear earlier: That's essentially what I was getting at; I was just pointing out that it would be nice to come up with a plot that exercises most of the configurable parameters in a stylesheet.

Generating all the figures shouldn't need to involve any Sphinx machinery: If you just plot multiple figures from a single script, they're rendered in the gallery as separate plots; e.g.

http://matplotlib.org/examples/color/colormaps_reference.html

CamDavidsonPilon · 2014-07-06T22:11:33Z

Kk, I've adding two scripts to generate pretty images using the new styles

tacaswell · 2014-07-06T22:14:17Z

examples/style_sheets/plot_fivethirtyeight.py

+
+x = np.linspace(0,10)
+
+figure(figsize=(6,4))


missing plt

adding np and plt into script

CamDavidsonPilon · 2014-07-08T12:17:59Z

bump

tonysyu · 2014-07-08T12:46:02Z

examples/style_sheets/plot_bmh.py

+plt.hist(beta(10,10,size=10000),histtype="stepfilled", bins=25, alpha=0.8, normed=True)
+plt.hist(beta(4,12,size=10000),histtype="stepfilled", bins=25, alpha=0.7, normed=True)
+plt.hist(beta(50,12,size=10000),histtype="stepfilled", bins=25, alpha=0.8, normed=True)
+plt.hist(beta(6,55,size=10000),histtype="stepfilled", bins=25, alpha=0.8, normed=True)


These lines raise some flake8 noise. Maybe you could define a small hist function to remove some repetition; e.g.

def step_hist(x): plt.hist(x, histtype="stepfilled", bins=25, alpha=0.8, normed=True)

Also, make sure there's a space after your commas.

tonysyu · 2014-07-08T14:32:12Z

Thanks for the quick updates, but I think it would be easier if you ran some sort of code checker. For example, flake8 reports the following on the plot_bmh.py script:

E302 expected 2 blank lines, found 1
E231 missing whitespace after ','
E231 missing whitespace after ','
E231 missing whitespace after ','
E231 missing whitespace after ','
E501 line too long (89 > 79 characters)
E231 missing whitespace after ','
E231 missing whitespace after ','
E231 missing whitespace after ','
E231 missing whitespace after ','

I know it seems a bit nitpicky, but many developers have code checkers connected to their editors that complain when there are PEP8 violations. BTW, I'm not a core developer, so I have no power to accept or reject this PR, so feel free to ignore me ;)

tacaswell · 2014-07-08T16:09:54Z

+1 on what Tony said (except ignoring him)

CamDavidsonPilon · 2014-07-09T00:07:02Z

hokay, I've pep'ed and flake'd both files

CamDavidsonPilon · 2014-07-11T00:55:28Z

thoughts, @tacaswell?

tonysyu · 2014-07-11T02:31:09Z

👍

tacaswell · 2014-07-11T04:41:01Z

I would like feed back from at least one more core dev and then it will get merged (most likely tomorrow).

tacaswell · 2014-07-12T16:23:23Z

As a side note (that I did not notice before) is that you should always create a feature branch to work on, rather that making new commits on master. The reason is that when this is merged you will need to use a force push to make your master on gh track the upstream master again.

Adding two new styles to mplstyles

CamDavidsonPilon · 2014-07-12T18:23:03Z

very true, thanks for informing me =)

Create 538.mplstyle

90d66db

CamDavidsonPilon reviewed Jul 5, 2014
View reviewed changes

Create bmh.mplstyle

bb29649

Update 538.mplstyle Update bmh.mplstyle fixing line in bmh.mplstyle

tacaswell added this to the v1.4.0 milestone Jul 6, 2014

tacaswell added needs_review and removed needs_review labels Jul 6, 2014

remove dpi/figsize settings, rename to fivethirtyeight

c2a5f5e

tacaswell reviewed Jul 6, 2014
View reviewed changes

examples/style_sheets/plot_fivethirtyeight.py

x = np.linspace(0,10)

figure(figsize=(6,4))

Copy link

Member

tacaswell Jul 6, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

missing plt

add plotting examples

2fba987

adding np and plt into script

tonysyu reviewed Jul 8, 2014
View reviewed changes

more pep8

7b70d19

tacaswell added a commit that referenced this pull request Jul 12, 2014

Merge pull request #3190 from CamDavidsonPilon/master

44b3634

Adding two new styles to mplstyles

tacaswell merged commit 44b3634 into matplotlib:master Jul 12, 2014

QuLogic removed the status: needs review label Nov 24, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding two new styles to mplstyles #3190

Adding two new styles to mplstyles #3190

CamDavidsonPilon commented Jul 5, 2014

tacaswell commented Jul 5, 2014

CamDavidsonPilon commented Jul 5, 2014

CamDavidsonPilon Jul 5, 2014

tacaswell commented Jul 6, 2014

tonysyu commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

Tillsten commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tacaswell commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tonysyu commented Jul 6, 2014

tonysyu commented Jul 6, 2014

tacaswell commented Jul 6, 2014

tonysyu commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tacaswell Jul 6, 2014

CamDavidsonPilon commented Jul 8, 2014

tonysyu Jul 8, 2014

tonysyu commented Jul 8, 2014

tacaswell commented Jul 8, 2014

CamDavidsonPilon commented Jul 9, 2014

CamDavidsonPilon commented Jul 11, 2014

tonysyu commented Jul 11, 2014

tacaswell commented Jul 11, 2014

tacaswell commented Jul 12, 2014

CamDavidsonPilon commented Jul 12, 2014

Adding two new styles to mplstyles #3190

Adding two new styles to mplstyles #3190

Conversation

CamDavidsonPilon commented Jul 5, 2014

tacaswell commented Jul 5, 2014

CamDavidsonPilon commented Jul 5, 2014

CamDavidsonPilon Jul 5, 2014

Choose a reason for hiding this comment

tacaswell commented Jul 6, 2014

tonysyu commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

Tillsten commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tacaswell commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tonysyu commented Jul 6, 2014

tonysyu commented Jul 6, 2014

tacaswell commented Jul 6, 2014

tonysyu commented Jul 6, 2014

CamDavidsonPilon commented Jul 6, 2014

tacaswell Jul 6, 2014

Choose a reason for hiding this comment

CamDavidsonPilon commented Jul 8, 2014

tonysyu Jul 8, 2014

Choose a reason for hiding this comment

tonysyu commented Jul 8, 2014

tacaswell commented Jul 8, 2014

CamDavidsonPilon commented Jul 9, 2014

CamDavidsonPilon commented Jul 11, 2014

tonysyu commented Jul 11, 2014

tacaswell commented Jul 11, 2014

tacaswell commented Jul 12, 2014

CamDavidsonPilon commented Jul 12, 2014