DOC improved example plot in plot_lda_qda.py #12942
Conversation
|
Thanks for the pr. In the image above, the two lower figures should look like identical datasets, but now don't |
|
they're the same, how not? It's just that misclassified data are marked with crosses. I haven't modified data generation process at all. |
|
Ah, you're right. I was confused visually by the ellipses which are too close in colour to the points |
|
Please increase the alpha on the ellipses or otherwise increase contrast,
and re post the rendering. Thanks
|
|
Ellipses are colored this to be consistent with what data class they're representing. Increasing alpha will obscure the data points even more, and I am not sure what you mean by contrast here. Honestly, I think this is good, much clearer compared to the original plot. The confusion may be because there is a lot of information to show on this plot, it may take a moment to figure out which is which. |
|
I meant decreasing alpha, sorry.
…On Wed, 9 Jan 2019 at 12:36, Zhuyi Xue ***@***.***> wrote:
Ellipses are colored this to be consistent with what data class they're
representing. Increasing alpha will obscure the data points even more, and
I am not sure what you mean by contrast here.
Honestly, I think this is good, much clearer compared to the original
plot. The confusion may be because there is a lot of information to show on
this plot, it may take a moment to figure out which is which.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#12942 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAEz6_BF5G9GTj8ysu45vfUyXa8t3aFJks5vBUe6gaJpZM4Z186o>
.
|
|
I could make it transparent, as seen here. Let me know if it's ok. If so, I will push the update. |
|
Looking at them on a big screen, I'm okay with what you've currently got.
Probably was just the cheap display on my mobile.
|
|
Should be good now. The current render: |
)" This reverts commit fc5c2a3.
)" This reverts commit fc5c2a3.
Improved the plots on https://scikit-learn.org/stable/auto_examples/classification/plot_lda_qda.html#sphx-glr-auto-examples-classification-plot-lda-qda-py.
Currently, the quadratic decision boundary of QDA is hardly seen in the 4th subplot. As the boundary being quadratic is the most important distinguishment from that of LDA, the current version could be confusing.
The updated plot: