add "statistics" category to quantiles

[soxofaan]

openeo-processes/quantiles.json

Lines 2 to 7 in 60fb5e2

	"id": "quantiles",
	"summary": "Quantiles",
	"description": "Calculates quantiles, which are cut points dividing the range of a probability distribution into either\n\n* intervals corresponding to the given `probabilities` or\n* (nearly) equal-sized intervals (q-quantiles based on the parameter `q`).\n\nEither the parameter `probabilities` or `q` must be specified, otherwise the `QuantilesParameterMissing` exception is thrown. If both parameters are set the `QuantilesParameterConflict` exception is thrown.",
	"categories": [
	"math"
	],

Strictly speaking,quantiles can not be used directly as a reducer (you need an additional array_element to get a scalar value from its output array), but I think it makes sense to add it to the "reducer" category because it will be used often in reduce contexts.

[m-mohr]

Hmm, I'm not sure I agree. This would make it less clear what a reducer strictly is. It would not just be quantiles, too, as there are more similar processes, e.g. extrema.

Looking at the reducers right now, I'd even argue listing the array_ functions is a bit strange and I'd consider removing the reducer category from them.

[soxofaan]

I guess it depends on what a "category" means to you,
if you see it as some kind of (data) type/schema or contract, then yes, quantiles is not a reducer (because it doesn't convert a collection to a single value). But then reduce_dimension or reduce_spatial should also not be listed there.

To me, the categories are just thematic labels to help with finding processes you need for a particular use case. In case of quantiles: because median (a particular quantile) is listed as reducer, I'd also expect to find other quantiles there.

Other approach: change category reducer to reduce (like form of category export) or reducing (like form of category sorting) to make it sound less strict.

[m-mohr]

Those are labels, but they should still be somewhat consistently chosen. So we'd need to add them not just to quantiles, but also to extrema. I'm just trying to avoid that someone simply tries to use quantiles exactly as e.g. mean and then it fails. The docs refer to reducers and say that you can use them, but in the quantiles/extrema/... cases it's not as simple...

[soxofaan]

But then reduce_dimension and reduce_spatial should also not be listed there.

Other solution: can't we also define a category "statistics" , to list these reduce/summary processes like median, mean, min, max, extrema, quantiles, ...
but not those array_ processes?

[m-mohr]

Yes, that sounds like a good solution, I think.

[soxofaan]

Meta question: can we label this ticket with something like "low-hanging-fruit" or "beginner-material" so that aspiring contributors can easily find simple issues to start working on.

e.g. for python client I started using "low hanging fruit": https://github.com/Open-EO/openeo-python-client/issues?q=is%3Aissue+is%3Aopen+label%3A%22low+hanging+fruit%22

[m-mohr]

I've re-added the "good first issue" label, which is common on GitHub.

[m-mohr]

PR is available for review: #294