[MRG + 1] Plot iso-f1 curves in plot_precision_recall example #8378

SACHIN-13 · 2017-02-16T19:44:48Z

Reference Issue

What does this implement/fix? Explain your changes.

While plotting PR curves, it separates out the curves with equal f1 values.

Any other comments?

An edit to my previous PR #8366

SACHIN-13 · 2017-02-16T19:48:50Z

I had removed the previous branch from my fork, so was unable to force push the changes in the same PR, my bad!
So, created a new PR with suggested changes!
Please review it.

codecov · 2017-02-16T23:58:31Z

Codecov Report

Merging #8378 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #8378   +/-   ##
=======================================
  Coverage   94.75%   94.75%           
=======================================
  Files         342      342           
  Lines       60801    60801           
=======================================
  Hits        57609    57609           
  Misses       3192     3192

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8694278...a1b4837. Read the comment docs.

SACHIN-13 · 2017-02-17T17:51:03Z

SACHIN-13 · 2017-02-19T10:15:41Z

Waiting for reviews!

ogrisel

Looks good to me, thank you for your contribution!

ogrisel · 2017-02-19T11:00:27Z

examples/model_selection/plot_precision_recall.py

+labels = []
+for F in f_scores:
+    x = R = np.linspace(0.01, 1)
+    y = F * R / (2 * R - F)


I would rather use lower case variable names for 1d vectors.

also you could use f_score instead of f maybe? And reuse x instead of R?

raghavrv

+1 from me once addressed...

raghavrv · 2017-02-19T12:17:23Z

examples/model_selection/plot_precision_recall.py

+labels = []
+for F in f_scores:
+    x = R = np.linspace(0.01, 1)
+    y = F * R / (2 * R - F)


also you could use f_score instead of f maybe? And reuse x instead of R?

raghavrv · 2017-02-19T12:20:13Z

~~How does it appear on the website? I can't see the artifacts of circle ci...~~ Ah got it (one needs to login duh!)

raghavrv · 2017-02-19T12:25:22Z

Can we have it (the legends) not overlapping on the curves? It looks a bit congested as such now...

Putting the legend outside the curve would be nice...

raghavrv · 2017-02-19T12:26:36Z

Also can you set an alpha param to control translucency on the iso-f1?

codecov-io · 2017-02-19T12:49:25Z

Codecov Report

Merging #8378 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #8378   +/-   ##
=======================================
  Coverage   94.75%   94.75%           
=======================================
  Files         342      342           
  Lines       60813    60813           
=======================================
  Hits        57621    57621           
  Misses       3192     3192

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1ba84d7...0374684. Read the comment docs.

SACHIN-13 · 2017-02-19T12:54:50Z

Yes, changes made according to the reviews! I set alpha=0.2.

jnothman · 2017-02-19T12:59:37Z

I think we should cut it back to 0.2, 0.4, 0.6, 0.8. Nine curves are too many

SACHIN-13 · 2017-02-19T13:00:42Z

okay! 👍

raghavrv

That's it looks good to me... Thanks! Will merge when the CIs pass...

SACHIN-13 · 2017-02-19T13:45:28Z

You're welcome :)

raghavrv · 2017-02-19T17:45:45Z

The legend is still inside the plot obstructing the view. Could you move it outside please?

raghavrv · 2017-02-19T17:46:20Z

Or increase the size of the plot such that it can fit into the enlarged blank space...

SACHIN-13 · 2017-02-19T21:05:48Z

does this look good?

raghavrv · 2017-02-19T21:37:37Z

Perfect. Thanks!

raghavrv

Also can you remove the plt.show above? it makes it not to display in the docs... (it's not a part of this pr actually...)

raghavrv · 2017-02-19T21:38:54Z

examples/model_selection/plot_precision_recall.py

+for f_score in f_scores:
+    x = np.linspace(0.01, 1)
+    y = f_score * x / (2 * x - f_score)
+    l, = plt.plot(x[y >= 0], y[y >= 0], color='gray', alpha=0.2)


unneeded , ?

Also can you directly do append here?

without the , , l becomes [<matplotlib.lines.Line2D object at 0x7f43d13912e8>] , which is not supported by legend (a UserWarning is raised). Using a , returns a line instance Line2D(_line3) which is what the legend function accepts! So, the lines have to be appended separately.

raghavrv · 2017-02-19T21:40:11Z

examples/model_selection/plot_precision_recall.py

-                   ''.format(i, average_precision[i]))
-
+    l, = plt.plot(recall[i], precision[i], color=color, lw=lw)
+    lines.append(l)


same comment as above...

same as above.

SACHIN-13 · 2017-02-20T02:55:37Z

The previous curve(pr curve for class 0) is already been included in the next plot(which is being displayed in the docs).

SACHIN-13 · 2017-02-20T02:56:03Z

Had missed a legend for the "micro-average precision curve". Added it!

ogrisel · 2017-02-20T12:53:35Z

I am still +1 for merging :)

ogrisel · 2017-02-20T12:53:46Z

Let's merge then.

ogrisel · 2017-02-20T12:54:03Z

Thanks @SACHIN-13 !

SACHIN-13 mentioned this pull request Feb 16, 2017

plot iso-f1 curves in plot_precision_recall #8366

Closed

ogrisel approved these changes Feb 19, 2017

View reviewed changes

raghavrv suggested changes Feb 19, 2017

View reviewed changes

SACHIN-13 force-pushed the my-feature branch from a1b4837 to 7dc3e06 Compare February 19, 2017 12:49

SACHIN-13 force-pushed the my-feature branch from 7dc3e06 to d30ed84 Compare February 19, 2017 13:17

raghavrv approved these changes Feb 19, 2017

View reviewed changes

SACHIN-13 changed the title ~~plot iso-f1 curves in plot_precision_recall~~ [MRG]plot iso-f1 curves in plot_precision_recall Feb 19, 2017

raghavrv changed the title ~~[MRG]plot iso-f1 curves in plot_precision_recall~~ [MRG] Plot iso-f1 curves in plot_precision_recall example Feb 19, 2017

raghavrv changed the title ~~[MRG] Plot iso-f1 curves in plot_precision_recall example~~ [MRG + 1] Plot iso-f1 curves in plot_precision_recall example Feb 19, 2017

SACHIN-13 force-pushed the my-feature branch from d30ed84 to f74d7d6 Compare February 19, 2017 21:05

raghavrv reviewed Feb 19, 2017

View reviewed changes

plot iso-f1 curves in plot_precision_recall

0374684

SACHIN-13 force-pushed the my-feature branch from f74d7d6 to 0374684 Compare February 20, 2017 02:56

ogrisel merged commit 0cae688 into scikit-learn:master Feb 20, 2017

sergeyf pushed a commit to sergeyf/scikit-learn that referenced this pull request Feb 28, 2017

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

674284f

Przemo10 mentioned this pull request Mar 17, 2017

update fork (#1) #8606

Closed

Sundrique pushed a commit to Sundrique/scikit-learn that referenced this pull request Jun 14, 2017

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

9a680c8

NelleV pushed a commit to NelleV/scikit-learn that referenced this pull request Aug 11, 2017

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

b943a33

paulha pushed a commit to paulha/scikit-learn that referenced this pull request Aug 19, 2017

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

a193648

maskani-moh pushed a commit to maskani-moh/scikit-learn that referenced this pull request Nov 15, 2017

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

b38986b

lemonlaug pushed a commit to lemonlaug/scikit-learn that referenced this pull request Jan 6, 2021

plot iso-f1 curves in plot_precision_recall (scikit-learn#8378)

b30751e

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MRG + 1] Plot iso-f1 curves in plot_precision_recall example #8378

[MRG + 1] Plot iso-f1 curves in plot_precision_recall example #8378

SACHIN-13 commented Feb 16, 2017

SACHIN-13 commented Feb 16, 2017 •

edited

Loading

codecov bot commented Feb 16, 2017

SACHIN-13 commented Feb 17, 2017

SACHIN-13 commented Feb 19, 2017

ogrisel left a comment

ogrisel Feb 19, 2017

raghavrv Feb 19, 2017

raghavrv left a comment

raghavrv Feb 19, 2017

raghavrv commented Feb 19, 2017 •

edited

Loading

raghavrv commented Feb 19, 2017 •

edited

Loading

raghavrv commented Feb 19, 2017

codecov-io commented Feb 19, 2017 •

edited

Loading

SACHIN-13 commented Feb 19, 2017

jnothman commented Feb 19, 2017

SACHIN-13 commented Feb 19, 2017

raghavrv left a comment

SACHIN-13 commented Feb 19, 2017

raghavrv commented Feb 19, 2017

raghavrv commented Feb 19, 2017

SACHIN-13 commented Feb 19, 2017

raghavrv commented Feb 19, 2017

raghavrv left a comment •

edited

Loading

raghavrv Feb 19, 2017

raghavrv Feb 19, 2017

SACHIN-13 Feb 20, 2017

raghavrv Feb 19, 2017

SACHIN-13 Feb 20, 2017

SACHIN-13 commented Feb 20, 2017

SACHIN-13 commented Feb 20, 2017 •

edited

Loading

ogrisel commented Feb 20, 2017

ogrisel commented Feb 20, 2017

ogrisel commented Feb 20, 2017

[MRG + 1] Plot iso-f1 curves in plot_precision_recall example #8378

[MRG + 1] Plot iso-f1 curves in plot_precision_recall example #8378

Conversation

SACHIN-13 commented Feb 16, 2017

Reference Issue

What does this implement/fix? Explain your changes.

Any other comments?

SACHIN-13 commented Feb 16, 2017 • edited Loading

codecov bot commented Feb 16, 2017

Codecov Report

SACHIN-13 commented Feb 17, 2017

SACHIN-13 commented Feb 19, 2017

ogrisel left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

raghavrv left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

raghavrv commented Feb 19, 2017 • edited Loading

raghavrv commented Feb 19, 2017 • edited Loading

raghavrv commented Feb 19, 2017

codecov-io commented Feb 19, 2017 • edited Loading

Codecov Report

SACHIN-13 commented Feb 19, 2017

jnothman commented Feb 19, 2017

SACHIN-13 commented Feb 19, 2017

raghavrv left a comment

Choose a reason for hiding this comment

SACHIN-13 commented Feb 19, 2017

raghavrv commented Feb 19, 2017

raghavrv commented Feb 19, 2017

SACHIN-13 commented Feb 19, 2017

raghavrv commented Feb 19, 2017

raghavrv left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SACHIN-13 commented Feb 20, 2017

SACHIN-13 commented Feb 20, 2017 • edited Loading

ogrisel commented Feb 20, 2017

ogrisel commented Feb 20, 2017

ogrisel commented Feb 20, 2017

SACHIN-13 commented Feb 16, 2017 •

edited

Loading

raghavrv commented Feb 19, 2017 •

edited

Loading

raghavrv commented Feb 19, 2017 •

edited

Loading

codecov-io commented Feb 19, 2017 •

edited

Loading

raghavrv left a comment •

edited

Loading

SACHIN-13 commented Feb 20, 2017 •

edited

Loading