Skip to content

Fix #130: Fallback to full range if IQR is zero in RobustScaler #131

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

Open
KelianM wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix #130: Fallback to full range if IQR is zero in RobustScaler #131

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 11 additions & 2 deletions gluon_utils/scalers/robust_scaler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@
class RobustScaler(Scaler):
"""
Computes a scaling factor by removing the median and scaling by the
interquartile range (IQR).
interquartile range (IQR) or, if IQR is 0, by the full range.

Parameters
----------
Expand Down Expand Up @@ -61,9 +61,18 @@ def __call__(
q3 = torch.nanquantile(observed_data, 0.75, dim=self.dim, keepdim=True)
iqr = q3 - q1

# Compute full range as a fallback if IQR is 0
data_min = torch.nanquantile(observed_data, 0, dim=self.dim, keepdim=True)
data_max = torch.nanquantile(observed_data, 1, dim=self.dim, keepdim=True)
full_range = data_max - data_min

# if observed data is all zeros, nanmedian returns nan
loc = torch.where(torch.isnan(med), torch.zeros_like(med), med)
scale = torch.where(torch.isnan(iqr), torch.ones_like(iqr), iqr)
scale = torch.where(
torch.isnan(iqr),
torch.ones_like(iqr),
torch.where(iqr == 0, full_range, iqr)
)
scale = torch.maximum(scale, torch.full_like(iqr, self.minimum_scale))

scaled_data = (data - loc) / scale
Expand Down