tedana.metrics.dependence.calculate_partial_r2

calculate_partial_r2(*, semipartial_r2: ndarray, total_r2: float) → ndarray

Calculate mean voxelwise partial R-squared for each regressor.

This is equivalent to the variance explained by each component after regressing the other components out of the data and the component itself. It is a conditional effect size.

We simplify the math by calculating partial R-squared as semi-partial R-squared / (semi-partial R-squared + (100 - total R-squared)).

Parameters:

• semipartial_r2 ((C) array_like) – Semi-partial R-squared for each component.

• total_r2 (float) – Total R-squared for all components.

Returns:

partial_r2 ((C) array_like) – Average (across voxels) partial R-squared for each regressor, on a scale from 0 to 100.