App pubfacility

[hupili]

As a case learning case for current map API.

One remaining problem is how to present NaN. Maybe it's better to make the are transparent?

[hxu]

I think transparent may work better for NaNs. NaNs should probably also be coerced to 0s, since for the public facility dataset, that's what they actually mean.

This is a pretty good start, but I think could use a bit of polish. There are some cases where there is some ambiguity -- for example, the dual dropdowns don't make it clear what indicator you're actually showing. I can guess that it was my last choice, but there's potential for confusion (maybe some people would think it's a combination of the two chosen indicators or something). Maybe we can forcibly clear the other dropdown when one is selected?

Another thing is that my quantization heuristic is not well suited for the public facilities use case. It should probably check the number of unique values in the data, and if it is low, then it should not quantize. With this public facility data, we get weird categories like "1-1" and such.

Would prefer to fix these issues before merging in.

[hupili]

@hxu Please feel free to modify the presentation. I simply wrap the UI to test backend and I agree with the above problems. Especially for quantisation, the public facility counts are skewed:

• most areas only have 0,1,2
• A few areas have extra large values.

There is no differentiation for small values. It would be better to differentiate small values.