pypomp.Pomp.traces

Pomp.traces() → DataFrame

Return the full trace of log-likelihoods and parameters over all runs.

Concatenates the parameter and log-likelihood histories from every pfilter(), mif(), and train() call stored in results_history.

Returns:

Long-format DataFrame containing concatenated trace data. The columns appear in the following order:

1. theta_idx: The index of the parameter set.

2. iteration: Counter indicating the global iteration.

3. method: The name of the method (e.g. 'pfilter', 'mif', 'train').

4. logLik: The estimated log-likelihood.

5. se: The standard error of the log-likelihood estimate.

6. Parameter columns: One column per model parameter in their defined order.

Return type:

pd.DataFrame