frites.estimator.ResamplingEstimator#

class frites.estimator.ResamplingEstimator(estimator, n_resampling=100, verbose=None)[source]#

Trial-resampling estimator.

In case of unbalanced contrast (i.e. when the number of trials per condition is very different) it can be interesting to use a trial-resampling technique to minimize the possibility that the effect size is driven by the number of trials. To this end, the same number of trials is used to estimate the effect size and the final

Parameters
estimatorfrites.estimator

An estimator object (e.g. GCMIEstimator, CorrEstimator etc.)

n_resamplingpython:int | 100

Number of resampling to perform

Methods

estimate(x, y[, z, categories])

Estimate the amount of information shared with resampling.

get_function()

Get the function to execute according to the input parameters.
estimate(x, y, z=None, categories=None)[source]#

Estimate the amount of information shared with resampling.

This method is made for estimating information on 3D variables (i.e (n_var, n_mv, n_samples)) where n_var is an additional dimension (e.g times, times x freqs etc.), n_mv is a multivariate axis and n_samples the number of samples. When computing MI, both the multivariate and samples axes are reduced.

In contrast to other estimators, the resampling estimator need to know whether the trials belong to the first or second condition. Therefore, a vector of discrete conditions need to be provided, depending of the type of information :

  • If mi_type is ‘cc’ : x and y should be continuous variables and z should contains the discrete classes

  • If mi_type is ‘cd’ : x should be continuous and the y variables is used to find the discrete classes

Parameters
xnumpy:array_like

Array of shape (n_var, n_mv, n_samples).

ynumpy:array_like

Array with a shape that depends on the type of MI (mi_type) :

  • If mi_type is ‘cc’, y should be an array with the same shape as x

  • If mi_type is ‘cd’, y should be a row vector of shape (n_samples,). In that case, the discrete values of y are used to identify the discrete classes

znumpy:array_like | python:None

Array containing discrete classes in case mi_type is ‘cc’. Should be a vector of shape (n_samples,)

categoriesnumpy:array_like | python:None

Row vector of categories. This vector should have a shape of (n_samples,) and should contains integers describing the category of each sample.

Returns
infonumpy:array_like

Array of information of shape (n_categories, n_var). If categories is None when computing information, n_categories is going to be one.

Examples using estimate:

Trial-resampling: correcting for unbalanced designs

Trial-resampling: correcting for unbalanced designs
get_function()[source]#

Get the function to execute according to the input parameters.

This can be particularly useful when computing resampling in parallel as it avoids to pickle the whole estimator and therefore, leading to faster computations.

The returned function has the following signature :

  • fcn(x, y, z=None, categories=None)

and return an array of shape (n_categories, n_var).

Examples using frites.estimator.ResamplingEstimator#

Trial-resampling: correcting for unbalanced designs

Trial-resampling: correcting for unbalanced designs