Simulate

utils.simulate.generate_sequence(config, labels_df, T, n_subjs=1, random_state=42, verbose=False)[source]

Generate individual simulated pose module label sequence of length T (noting that T = number of observations, not time)

Parameters:

  • config – the config object

  • labels_df – labels_df from analyze.get_module_labels

  • T – length of sequence to be generated

  • n_subjs – number of subjects to generate (if labels_df has subgroups, will be number of subjects PER GROUP)

  • random_state – random state seed (default: 42)

  • verbose – print progress (default: False)

Returns:

sequence

utils.simulate.generate_usage(module_feature_object, n_samples, random_state=42, mode='log-normal', scale=10)[source]

Generate simulated pose module usage object

Parameters:

  • module_feature_object – module feature object of class ModuleUsage (from analyze.get_module_usage) or ModuleTransitions (from analyze.get_module_transitions)

  • n_samples – number of samples to generate

  • random_state – random state seed (default: 42)

  • mode – ‘log-normal’ or ‘multivariate_gaussian’; default log-normal

  • scale – scale factor for variance in log-normal mode (not needed for multivariate_gaussian mode)

Returns:

module_feature_object of the same style as the one input

utils.simulate.generate_usage_labeled(module_feature_object, n_samples_per_bin, bins, regression, max_iters='default', random_state=42, mode='log-normal', scale=10, verbosity='medium')[source]

Generate simulated pose module usage object

Parameters:

  • module_feature_object – module feature object of class ModuleUsage (from analyze.get_module_usage) or ModuleTransitions (from analyze.get_module_transitions)

  • n_samples_per_bin – number of samples to generate per bin in timebin

  • bins – 2D array of bins containing upper and lower limit for each bin

  • regression – regression model to label samples

  • max_iters – number for how many iterations to attempt to generate samples; will raise an error if exceeded without generating enough samples; number or ‘default’ for n_samples_total*10

  • random_state – random state seed (default: 42)

  • mode – ‘log-normal’ or ‘multivariate_gaussian’; default log-normal

  • scale – scaling factor for covariance in log-transformed approach

  • verbosity – ‘low’,’medium’, or ‘high’

Returns:

module_feature_object of the same style as the one input