NIPY logo

Site Navigation

NIPY Community

Github repo

Table Of Contents

Next topic

labs.group.permutation_test

This Page

labs.group.mixed_effects

Module: labs.group.mixed_effects

New generic implementation of multiple regression analysis under noisy measurements.

Functions

nipy.labs.group.mixed_effects.em(Y, VY, X, C=None, niter=2, log_likelihood=False)

Maximum likelihood regression in a mixed-effect linear model using the EM algorithm.

Parameters :

Y : array

Array of observations.

VY : array

C is the contrast matrix. Conventionally, C is p x q where p :

is the number of regressors. :

OUTPUT: beta, s2 :

beta – array of parameter estimates :

s2 – array of squared scale parameters. :

REFERENCE: :

Keller and Roche, ISBI 2008. :

nipy.labs.group.mixed_effects.log_likelihood_ratio(Y, VY, X, C, niter=2)

Log-likelihood ratio statistic: 2*(log L - log L0)

It is asymptotically distributed like a chi-square with rank(C) degrees of freedom under the null hypothesis H0: Cb = 0.

nipy.labs.group.mixed_effects.nonzero(x)

Force strictly positive values.