fix(regression): return variance (σ²) instead of std dev (σ) in DummyProbaRegressor._predict_var

[tysoncung]

Problem

DummyProbaRegressor._predict_var returns the standard deviation (σ) of the training labels instead of the variance (σ²). This is a unit-mismatch bug — predict_var should return variance, not standard deviation.

Solution

Changed self._sigma to self._sigma ** 2 in _predict_var, so the method correctly returns variance (σ²) as documented.

Files Changed

• skpro/regression/dummy.py: Fixed _predict_var to return self._sigma ** 2 instead of self._sigma

Verification

import numpy as np
# Before fix: _predict_var returns np.std(y) ≈ σ
# After fix: _predict_var returns np.std(y) ** 2 ≈ σ² (variance)

Closes #975