shape_index is the same as radii_ratio

[martinfleis]

We have a duplication in shape indices. shape_index is the same as radii_ratio.

esda/esda/shape.py

Line 356 in 17c5f00

return numpy.sqrt(shapely.area(ga) / numpy.pi) / shapely.minimum_bounding_radius(ga)

and

esda/esda/shape.py

Lines 193 to 195 in 17c5f00

	r_eac = numpy.sqrt(shapely.area(ga) / numpy.pi)
	r_mbc = shapely.minimum_bounding_radius(ga)
	return r_eac / r_mbc

are the same formulas.

We should probably deprecate one in favour of the other.

[ljwolf]

I like the “radii_ratio” name, since it expresses its structure as the ratio of two ideal shape radii. The two implementations are basically the same, though I find the one with more lines to be easier to understand.

We can always alias one as the other?

[martinfleis]

Yeah, I also like the radii ratio better as shape index is too generic. All of them are shape indices. Agree that the multiline version is a bit more readable.

We can always alias one as the other?

Yeah but I don't think there's a ton of downstream usage of this so I would simply deprecate shape_index and add a note to radii_ratio docstring saying that it is also known as Schumm's shape index.

[ljwolf]

I will never pass up an opportunity to deprecate!

[martinfleis]

@leehach in light of our recent discussions on naming (and #408), what would be your suggestion here? It is clear that we don't need both and it is also likely that "shape index" is not a good name (of any metric...). Just want to make sure we do this once.

[leehach]

I agree. "Shape index" is not adequately informative, as these various measures are all shape indices of some form or another. I agree with retaining radii_ratio().