Framework#

The framework module specifies the operation of the DCMR model using the CyMR engine. Model variants may be configured using model_variant(). Model specifications are defined by WeightParameters objects, which may be stored in JSON files.

This module also provides functions for running parameter searches, evaluating models using cross-validation, and working with fit and cross-validation results.

Model configuration#

model_variant(fcf_features[, ff_features, ...])

Define parameters for a model variant.

Evaluating models#

`run_fit`(res_dir, data, param_def, patterns, ...)	Fit parameters to individual subjects of a dataset.
`run_xval`(res_dir, data, param_def, patterns)	Evaluate a model using cross-validation.

Fit results#

`read_fit_param`(fit_file)	Read subject parameters from a fit results file.
`read_fit_weights`(param_file)	Read weights from a parameters file.
`read_model_spec`(def_file)	Read model specification file as a series.
`read_model_specs`(fit_dir, models[, model_names])	Read model definitions for multiple models.
`read_model_fits`(fit_dir, models[, ...])	Read fit results for multiple models.
`read_model_xvals`(fit_dir, models[, model_names])	Read cross-validation results for multiple models.
`read_model_sims`(data_file, fit_dir, models)	Read simulated data for multiple models.
`get_sim_models`(study, model_set[, included])	Get a list of models for a study.
`compare_fit`(means)	Compare model fit to mean measures.

Model parameters#

`WeightParameters`()	Manage CFR parameters.
`WeightParameters.set_scaling_param`(...[, ...])	Add scaling parameters for patterns or similarity.
`WeightParameters.set_intercept_param`(...)	Set a free intercept parameter for each connection matrix.
`WeightParameters.set_region_weights`(connect, ...)	Set weights within context regions.
`WeightParameters.set_sublayer_weights`(...[, ...])	Set weights for context sublayers.
`WeightParameters.set_item_weights`(...[, ...])	Set item-item weights.
`WeightParameters.set_learning_sublayer_param`(...)	Set dependent sublayer parameters for learning.
`WeightParameters.set_weight_sublayer_param`(...)	Set scaling of sublayer learning rates.
`WeightParameters.set_free_sublayer_param`(...)	Set sublayer parameters to be free.