dynode.utils.vis_utils.plot_checkpoint_inference_correlation_pairs#
- dynode.utils.vis_utils.plot_checkpoint_inference_correlation_pairs(posteriors_in: dict[str, ndarray | list], max_samples_calculated: int = 100, matplotlib_style: list[str] | str = ['seaborn-v0_8-colorblind'])#
Plot correlation pairs of sampled parameters with histograms and trend lines.
Parameters#
- posteriors_indict[str, np.ndarray | list]
Dictionary mapping parameter names to their posterior samples (shape: num_chains, num_samples). Parameters generated with numpyro.plate are flattened and displayed as separate parameters with _i suffix for each i in num_plates.
- max_samples_calculatedint
Maximum number of posterior samples per chain for calculations such as correlations and plotting. Set to -1 to disable cap; default is 100.
- matplotlib_stylelist[str] | str
Matplotlib style to use for plotting; default is [“seaborn-v0_8-colorblind”].
Returns#
- plt.Figure
Figure with n rows and n columns where n is the number of sampled parameters.