Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

DM-41814: Fix quantile scaling when number of bins is capped. #69

Merged
merged 2 commits into from
Nov 27, 2023
Merged

DM-41814: Fix quantile scaling when number of bins is capped. #69

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
47f8a4f
Add scaling on cdf to account for possible clipping.
erykoff Nov 22, 2023
29a2607
Add test for large range getQuantiles.
erykoff Nov 27, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
7 changes: 5 additions & 2 deletions python/lsst/summit/utils/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1011,8 +1011,10 @@ def getQuantiles(data, nColors):
if np.isnan(minVal): # cdf calculation has failed because all data is nan
return np.asarray([np.nan for _ in range(nColors)])

scale = (maxVal - minVal)/len(cdf)

boundaries = np.asarray(
[np.argmax(cdf >= i) + minVal for i in range(nColors)] + [maxVal]
[np.argmax(cdf >= i)*scale + minVal for i in range(nColors)] + [maxVal]
)
return boundaries

Expand All @@ -1034,5 +1036,6 @@ def digitizeData(data, nColors=256):
Scaled data in the [0, nColors - 1] range.
"""
cdf, minVal, maxVal = getCdf(data, nColors - 1)
bins = np.floor((data - minVal)).astype(np.int64)
scale = (maxVal - minVal)/len(cdf)
bins = np.floor((data*scale - minVal)).astype(np.int64)
return cdf[bins]
3 changes: 1 addition & 2 deletions tests/test_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,8 +233,7 @@ def test_quantiles(self):
# We understand that our algorithm gives very large rounding error
# compared to the generic numpy method. But still test it.
np.random.seed(1234)
# too big of a width violates the tolerance in the test to cap at 10k
dataRanges = [(50, 1, -1), (100_000, 5_000, -1), (5_000_000, 10_000, -2)]
dataRanges = [(50, 1, -1), (100_000, 5_000, -2), (5_000_000, 10_000, -2), (50_000, 100_000, -3)]
colorRanges = [2, 256, 999] # [very few, nominal, lots and an odd number]
for nColors, (mean, width, decimal) in itertools.product(colorRanges, dataRanges):
data = np.random.normal(mean, width, (100, 100))
Expand Down