dyngpt.tl.compute_kl_stats

dyngpt.tl.compute_kl_stats(result_infer)

Calculate statistical metrics and the Kullback-Leibler divergence between neural network-generated samples

and observed data.

Args:

result_infer (dict): A dictionary containing inference results with the following keys:

• “nn_sample_datas” (list or ndarray): Samples generated by the neural network.

• “observed_datas” (list or ndarray): Observed real-world sample data.

• “state_name” (list of str): List of state variable names.

• “model” (str): Name of the model used for inference.

Returns:

dict: A dictionary containing computed statistical metrics and KL divergence:

• “ssa_sample” (ndarray): Observed sample data.

• “nn_sample” (ndarray): Neural network-generated sample data.

• “mean_val_ssa” (pd.DataFrame): Mean values of observed data with state names as columns.

• “std_val_ssa” (pd.DataFrame): Standard deviations of observed data.

• “mean_val_nn” (pd.DataFrame): Mean values of neural network samples.

• “std_val_nn” (pd.DataFrame): Standard deviations of neural network samples.

• “kl_div” (pd.DataFrame): KL divergence values for each state variable.

• “model_name” (str): Model name used for inference.