Skip to content

Add support for sample_weights in DRTester #1005

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
Open
wants to merge 25 commits into
base: main
Choose a base branch
from
Open

Add support for sample_weights in DRTester #1005

wants to merge 25 commits into from

Conversation

gdaiha
Copy link
Contributor

@gdaiha gdaiha commented Sep 10, 2025

To solve issue #1004 , this PR adds support for sample_weights in DRTester class.
It implements it in fit_nuisances and in each evaluation (calibration, uplift and blp).

@gdaiha
added support for sample_weight in drtester
bed3194 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
changed numpy typing in econml results
20cadb8 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha gdaiha marked this pull request as draft September 10, 2025 21:13
@gdaiha
Copy link
Contributor Author

gdaiha commented Sep 10, 2025

I'm going to implement tests to this new feature

gdaiha and others added 22 commits September 10, 2025 18:18
@gdaiha
fixed sample_weight pass in fit_nuisance_train on validation sample
0970a65 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
implemented weighted quantiles and percentiles to avoid dependency on…
07e4d61 
… numpy>1.22

Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
created weighted utils module inside econml.validate
d2465e8 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
fixed weighted_stat module and standard error calculations in dr_tester
871f551 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
fixed standard error calculations in uplift
5b7ea4f 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
fixed n_size_bootstrap in bootstraping toc and qini curves
741a8e9 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
adjusted weighted averages in drtester
da4902e 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
removed weighted se in weighted_utils
7121acd 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
adjusted standard error calculation in calc_uplift
478043d 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
readded weighted_se and applied in drtester and calc_uplift
8d45b2b 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
Merge branch 'main' into gdaiha/add-sample-weights-drtester-signed
3e0cdc4
@gdaiha
standardized drtester vals_ to two dimensions
b2f006c 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
Merge branch 'gdaiha/add-sample-weights-drtester-signed' of https://g…
8be98c7 
…ithub.com/gdaiha/EconML into gdaiha/add-sample-weights-drtester-signed
@gdaiha
fixed treatments in drtester to allow controls different than zero
d87422b 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
fix assign units to validation sets in evaluate_cal to avoid same val…
ebe5074 
…ue in two groups

Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
changed pvalues to be two-sided in evaluate_uplift
e3c5ee3 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
added cross_val_predict_with_weights to drtester
966d8ce 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
guarantee support for cross_val_predict_with_weights to sklearn>=1.0
de12728 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
written tests for sample_weights support in drtester
f67239b 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
removed cast sample_weight to integer and changed to a normalization …
d10b57d 
…by minimum value

Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
cast weights vector to float to avoid typing error in numpy
28544e2 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha
cast weights vector to float to avoid typing error in numpy
c1e08dd 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
@gdaiha gdaiha marked this pull request as ready for review September 22, 2025 12:27
@gdaiha
Copy link
Contributor Author

gdaiha commented Sep 22, 2025

pr is ready to review. tests implemented.
@kbattocchi , could you rerun the two tests failed due to a temporary unavailability of uv?

@gdaiha
added comment in treatment loop
f910d52 
Signed-off-by: Gabriel Daiha <gabriel.alves@picpay.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@gdaiha