Fix #848: Pass estimation_sample_size parameter to individual trees in UpliftRandomForestClassifier

[mohsinm-dev]

Summary

Fixes issue #848 where the estimation_sample_size parameter was not being passed to individual trees in UpliftRandomForestClassifier.

Changes

• Added parameter storage in UpliftRandomForestClassifier.__init__ (line 2430)
• Added parameter passing to individual UpliftTreeClassifier instances during forest creation (line 2481)

Root Cause

The constructor accepted the estimation_sample_size parameter but failed to:

1. Store it as an instance variable (self.estimation_sample_size = estimation_sample_size)
2. Pass it to individual UpliftTreeClassifier instances during forest creation

Impact

This fix enables users to properly control honest splitting behavior in random forests when honesty=True. Previously, all trees defaulted to 0.5 regardless of user specification.

Testing

• Verified existing test suite passes (102/103 tests pass, 1 unrelated GraphViz failure)
• Confirmed no breaking changes or regressions
• Tested with multiple estimation_sample_size values (0.1-0.9)
• Verified default behavior preservation (0.5 when not specified)

Files Changed

• causalml/inference/tree/uplift.pyx (2 line addition)

[Copilot]

Pull Request Overview

This PR fixes a bug where the estimation_sample_size parameter was not properly passed to individual trees in UpliftRandomForestClassifier. The fix enables correct honest splitting behavior when honesty=True.

• Stores the estimation_sample_size parameter as an instance variable
• Passes the parameter to individual UpliftTreeClassifier instances during forest creation

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[jeongyoonlee]

LGTM. Thanks for your contribution, @mohsinm-dev!