Developer documentation
Version 3.0.3-105-gd3941f44
MR::Math::Stats::GLM::TestVariableHeteroscedastic Class Reference

#include "math/stats/glm.h"

Inheritance diagram for MR::Math::Stats::GLM::TestVariableHeteroscedastic:
MR::Math::Stats::GLM::TestVariableHomoscedastic MR::Math::Stats::GLM::TestBase

Protected Member Functions

void apply_mask_VG (const BitSet &mask, index_array_type &VG_masked, index_array_type &VG_counts) const
- Protected Member Functions inherited from MR::Math::Stats::GLM::TestVariableHomoscedastic
void get_mask (const size_t ie, BitSet &, const matrix_type &extra_columns) const
void apply_mask (const BitSet &mask, matrix_type::ConstColXpr data, const matrix_type &shuffling_matrix, const matrix_type &extra_column_data, matrix_type &Mfull_masked, matrix_type &shuffling_matrix_masked, vector_type &y_masked) const

Protected Attributes

const index_array_typeVG
const size_t num_vgs
vector_type gamma_weights
- Protected Attributes inherited from MR::Math::Stats::GLM::TestVariableHomoscedastic
const vector< CohortDataImport > & importers
const bool nans_in_data
const bool nans_in_columns
- Protected Attributes inherited from MR::Math::Stats::GLM::TestBase
const matrix_typey
const matrix_type M
const vector< Hypothesis > & c
std::shared_ptr< Math::Zstatisticstat2z

Detailed Description

A class to compute statistics from heteroscedastic data using a variable General Linear Model. This class produces a statistic per effect of interest. It should be used in cases where:

  • Additional subject data must be imported into the design matrix before computing t- / F-values; the design matrix therefore does not remain fixed for all elements being tested, but varies depending on the particular element being tested. How additional data is imported into the design matrix will depend on the particular type of data being tested. Therefore an Importer class must be defined that is responsible for acquiring and vectorising these data.
  • The input data are considered to be heteroscedastic; that is, the variance is not equivalent between all observations, but these can be placed into "variance groups", within which all observations can be considered to have the same variance.

Definition at line 427 of file glm.h.

The documentation for this class was generated from the following file: