Closed knaaptime closed 8 months ago
Merging #274 (2f557f2) into main (f6a8732) will increase coverage by
0.2%
. The diff coverage is97.4%
.
@@ Coverage Diff @@
## main #274 +/- ##
=======================================
+ Coverage 73.4% 73.6% +0.2%
=======================================
Files 24 24
Lines 3285 3316 +31
Branches 518 520 +2
=======================================
+ Hits 2411 2441 +30
Misses 708 708
- Partials 166 167 +1
Files | Coverage Δ | |
---|---|---|
esda/moran.py | 76.0% <97.4%> (+1.5%) |
:arrow_up: |
actually, it might be nice to just expose a method that returns cluster labels at a given threshold
How do you like to do the folium tests @martinfleis ?
Get html string and look for expected strings in that. Check geopandas explore tests, there are some helper functions to get that out of the m
.
I'll have a look at the PR later. Ping me if you'll need some input sooner.
appreciate it. It's coming back to me now as i'm remembering i already asked you for the local knox plots :P
this is ready, though not sure why the non-numba tests are failing
Seem to be failing due to dtype mismatch:
E Mismatched elements: 1 / 10 (10%)
E x: array(['Insignificant', 'Insignificant', 'Insignificant', 'Insignificant',
E 'Insignificant', 'Insignificant', 'Insignificant', 'Insignificant',
E 'Insignificant', 'Insignificant'], dtype=object)
E y: array(['Insignificant', 'Insignificant', 'Insignificant', 'Insignificant',
E 'Insignificant', 'Insignificant', 'High-High', 'Insignificant',
E 'Insignificant', 'Insignificant'], dtype='<U13')
@jGaboardi yeah thats weird but the vals are different too. the y array has a high-high
@jGaboardi yeah thats weird but the vals are different too. the y array has a high-high
good point; missed that part.
my guess would be something to do with the seeding
@martinfleis that would be great if you have a sec, but no sweat if not. I dunno why these are failing. Also if you think we should beef up the second map test a bit instead of just checking number of layers
I meant the actual coverage of the results, to check the html string for colours. Haven't actually looked at failures. But won't happen sooner than tomorrow eve.
yeah that would be great too. Though i think coverage should be decent once we're passing again
I have added a small commit checking the correctness of the folium html.
It seems that the result is stable on larger data. Maybe use Sacramento to test labels as well?
yeah, i'll just do that. Upping the number of permutations might do the trick too. Though its a little curious that all the numba-based tests are passing but the non-numba arent. I'm not sure what that implies about the seeding