Datasource objects (DS) take binned data and labels as inputs and generate population vectors (which are used by a cross-validator object to train and test a classifier). Datasources must implement a get_data method that returns the training and test data splits. The get_data method must have the following structure:
[all_XTr, all_YTr, all_XTe, all_YTe, ADDITIONAL_INFO] = ds.get_data
The outputs of this methods are:
1. all_XTr{iTime}{iCV} = [num_features x num_training_points]
A cell array that has the training data for all times and cross-validation splits
2. all_YTr = [num_training_point x 1]
The training labels
3. all_XTe{iTime}{iCV} = [num_features x num_test_points]
A cell array that has the test data for all times and cross-validation splits
4. all_YTe = [num_test_point x 1]
The test labels
The Neural Decoding Toolbox (version 1.0.0) comes with following datasource objects: