icc_analysis

Note

This page is a reference documentation. It only explains the function signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

icc_analysis(table: FeatureTable, repeat_tables: Sequence[FeatureTable], icc_types: Tuple[str, ...] = ('icc2', 'icc3'), min_subjects: int = 2, verbose: bool = False) → DataFrame[source]

Compute per-feature test-retest ICCs across measurement sessions.

This is the v1 successor of the v0.1 habit.core.machine_learning.feature_selectors.icc analysis: where that one merged CSVs from disk, this one consumes aligned feature tables directly and computes each ICC with the L0 kernels. Both the ICC(2,1) (two-way random, absolute agreement) and ICC(3,1) (two-way mixed, consistency) variants are reported by default, matching the v0.1 reporting defaults.

Parameters:

  • table – Primary-measurement table; its feature columns and row order define the analysis.

  • repeat_tables – One table per repeat measurement session, aligned to table by the identifier columns.

  • icc_types – Which ICC variants to compute ("icc2" and/or "icc3"); one result column per variant.

  • min_subjects – Minimum number of complete (NaN-free) subjects for a feature’s ICC to be computed; fewer yields NaN.

  • verbose – Show a progress bar over the features.

Returns:

Frame with one row per feature and the columns feature plus one column per requested ICC variant (NaN where undefined).

Raises:

HABITAPIError – If repeat_tables is empty or an unknown ICC variant is requested.