This looks good - although I wonder if now we risk people confusing the maximum value by the maximum index in each axis... Maybe others have an opinion?

@melissawm I know there was some more discussion in #18524 about this image and other images in the Creating Matrices section earlier this year. For this fix, I focused on what I could understand just in the aggregation example, but maybe the section as a whole needs a re-vamp? EDIT: Actually the changes requested have already been made in #18524, so I think this image is the only change left.

This looks good - although I wonder if now we risk people confusing the maximum value by the maximum index in each axis...

I agree @melissawm , I think your analysis+suggestion in this comment is spot on. As I see it, there are two options:

1. Update the first three images in the Creating matrices section (i.e. the two where the squares are blue and the one immediately following) and the code to use the data = np.array([[1, 2], [5, 3], [4, 6]]) array throughout as you've suggested, or
2. Modify the data prior to the portion of the example illustrating the reductions w/ axis to match the figure.

The latter option might not flow as nicely, but it has the advantage of not requiring any changes to the existing static figures. I've pushed up a concrete example of what that might look like in #20115.

I think after the discussion and seeing @rossbar 's proposal, I'd vote for closing this and just fixing the code. However, you mentioned there was trouble finding the specific fonts to change this, maybe the way to re-create these images could be documented somehow?

@MarsBarLee I'm closing this in favor of #20115. Thanks for providing this option.