Refactor univariate drift detection: centralize functionality in UnivariateDriftCalculator class

[Copilot]

Summary

This PR refactors the univariate drift detection module to centralize all functionality within the UnivariateDriftCalculator class while maintaining full backward compatibility. Previously, the codebase had both a class and several standalone functions with duplicated logic.

Changes Made

Centralized Logic in Class

• Moved normalize_wasserstein logic into UnivariateDriftCalculator.normalize_wasserstein() (static method)
• Moved detect_univariate_drift_with_options logic into UnivariateDriftCalculator.detect_drift_with_options() (static method)
• Added detect_drift() class method providing the functionality of detect_univariate_drift
• Added detect_drift_df() class method providing the functionality of detect_univariate_drift_df

Backward Compatibility

All standalone functions remain available and now delegate to the class methods:

# Old way still works
result = detect_univariate_drift_with_options(ref, cur, kind="continuous")

# New way using class methods
result = UnivariateDriftCalculator.detect_drift_with_options(ref, cur, kind="continuous")

# Both return identical results
assert result == result  # ✅

Internal Improvements

• UnivariateDriftCalculator.__call__() now uses its own detect_drift_with_options() method instead of calling the standalone function
• Eliminated code duplication between standalone functions and class
• Improved OOP design with proper encapsulation

Benefits

• ✅ Improved maintainability: All drift detection logic centralized in one class
• ✅ Better cohesion: Following OOP principles with related functionality grouped together
• ✅ Easier extensibility: Future drift detection features can be added as class methods
• ✅ Full backward compatibility: All existing code continues to work unchanged
• ✅ No breaking changes: All 138 tests pass, including 22 univariate drift tests

Example Usage

import pandas as pd
from tab_right.drift.univariate import UnivariateDriftCalculator

# Create test data
ref_df = pd.DataFrame({'numeric': [1, 2, 3], 'categorical': ['A', 'B', 'C']})
cur_df = pd.DataFrame({'numeric': [2, 3, 4], 'categorical': ['A', 'B', 'D']})

# New centralized approach
calc = UnivariateDriftCalculator(ref_df, cur_df)
result = calc()

# Or use static methods directly
result = UnivariateDriftCalculator.detect_drift_with_options(
    ref_df['numeric'], cur_df['numeric'], kind='continuous'
)

# Legacy functions still work exactly the same
from tab_right.drift.univariate import detect_univariate_drift_with_options
legacy_result = detect_univariate_drift_with_options(
    ref_df['numeric'], cur_df['numeric'], kind='continuous'
)

Fixes #118.

---

💬 Share your feedback on Copilot coding agent for the chance to win a $200 gift card! Click here to start the survey.