Coefficient of multiple correlation for repeated waveforms (Kadaba CMC)

The waveform analogue of a reliability coefficient (Kadaba et al. 1989): how similar a set of repeated curves are, relative to the curve's own variation. With groups = NULL the rows of x are repeated trials and the result is the within-day repeatability; pass groups (e.g. a session/day label per row) for the between-day CMC of the session-mean waveforms.

Usage

waveformCMC(x, groups = NULL)

Arguments

x: A trials x frames numeric matrix: each row one waveform (e.g. a joint angle over the time-normalised cycle), columns the cycle points.
groups: Optional grouping (length nrow(x)); when given, the CMC is computed between the group-mean waveforms (between-day/-condition).

Value

The CMC in [0, 1] (higher = more repeatable), or NA with a warning when the radicand is negative (waveform variation too small for a defined CMC – a known limitation for low-excursion curves).

References

Kadaba MP, et al. (1989) J Orthop Res 7:849-860.

Examples

base <- sin(seq(0, 2 * pi, length.out = 101))
trials <- t(sapply(1:5, function(i) base + rnorm(101, 0, 0.02)))
waveformCMC(trials)                                  # ~ 1 (repeatable)
#> [1] 0.9995692