elicit.plots module

elicit.plots.initialization(eliobj, cols: int = 4, **kwargs) → None

plots the ecdf of the initialization distribution per hyperparameter

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

colsint, optional

number of columns for arranging the subplots in the figure. The default is 4.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘init_matrix’ in eliobj.results. Have you excluded it from saving?

ValueError

if eliobj.results[“init_matrix”] is None: No samples from initialization distribution found. This plot function cannot be used if initial values were fixed by the user through the hyperparams argument in elicit.elicit.initializer().

Examples

>>> el.plots.initialization(eliobj, cols=6)

>>> el.plots.initialization(eliobj, cols=4, figsize=(8,3))

elicit.plots.loss(eliobj, weighted: bool = True, **kwargs) → None

plots the total loss and the loss per component.

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘loss_component’ in ‘eliobj.history’. Have you excluded ‘loss_components’ from history savings?

Can’t find ‘loss’ in ‘eliobj.history’. Have you excluded ‘loss’ from history savings?

Can’t find ‘elicited_statistics’ in ‘eliobj.results’. Have you excluded ‘elicited_statistics’ from results savings?

Examples

>>> el.plots.loss(eliobj, figsize=(8,3))

elicit.plots.hyperparameter(eliobj, cols: int = 4, **kwargs) → None

plots the convergence of each hyperparameter across epochs.

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

colsint, optional

number of columns for arranging the subplots in the figure. The default is 4.

spanint, optional

number of last epochs used to get a final averaged hyperparameter value. The default is 30.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘hyperparameter’ in ‘eliobj.history’. Have you excluded ‘hyperparameter’ from history savings?

Examples

>>> el.plots.hyperparameter(eliobj, figuresize=(8,3))

elicit.plots.prior_joint(eliobj, idx: int | list | None = None, **kwargs) → None

plot learned prior distributions of each model parameter based on prior samples from last epoch. If parallelization has been used, select which replication you want to investigate by indexing it through the ‘idx’ argument.

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

idxint or list or None:

only required if parallelization is used for fitting the method. Indexes the replications and allows to choose for which replication(s) the joint prior should be shown.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

ValueError

Currently only ‘positive’ can be used as constraint. Found unsupported constraint type.

The value for ‘idx’ is larger than the number of parallelizations.

KeyError

Can’t find ‘prior_samples’ in ‘eliobj.results’. Have you excluded ‘prior_samples’ from results savings?

Examples

>>> el.plots.prior_joint(eliobj, figsize=(4,4))

elicit.plots.prior_marginals(eliobj, cols: int = 4, **kwargs) → None

plots the convergence of each hyperparameter across epochs.

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

colsint, optional

number of columns for arranging the subplots in the figure. The default is 4.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘prior_samples’ in ‘eliobj.results’. Have you excluded ‘prior_samples’ from results savings?

Examples

>>> el.plots.prior_marginals(eliobj, figuresize=(8,3))

elicit.plots.elicits(eliobj, cols: int = 4, **kwargs) → None

plots the expert-elicited vs. model-simulated statistics.

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

colsint, optional

number of columns for arranging the subplots in the figure. The default is 4.

**kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘expert_elicited_statistics’ in ‘eliobj.results’. Have you excluded ‘expert_elicited_statistics’ from results savings?

Can’t find ‘elicited_statistics’ in ‘eliobj.results’. Have you excluded ‘elicited_statistics’ from results savings?

Examples

>>> el.plots.elicits(eliobj, cols=4, figsize=(7,3))

elicit.plots.marginals(eliobj, cols: int = 4, span: int = 30, **kwargs) → None

plots convergence of mean and sd of the prior marginals

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

colsint, optional

number of columns for arranging the subplots in the figure. The default is 4.

spanint, optional

number of last epochs used to get a final averaged value for mean and sd of the prior marginal. The default is 30.

kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘hyperparameter’ in ‘eliobj.history’. Have you excluded ‘hyperparameter’ from history savings?

Examples

>>> el.plots.marginals(eliobj, figuresize=(8,3))

elicit.plots.priorpredictive(eliobj, **kwargs) → None

plots prior predictive distribution of samples from the generative model in the last epoch

Parameters:

eliobjinstance of elicit.elicit.Elicit()

fitted eliobj object.

kwargsany, optional

additional keyword arguments that can be passed to specify plt.subplots()

Raises:

KeyError

Can’t find ‘target_quantities’ in ‘eliobj.results’. Have you excluded ‘target_quantities’ from results savings?

Examples

>>> el.plots.priorpredictive(eliobj, figuresize=(6,2))

elicit.plots.prior_averaging(eliobj, cols: int = 4, n_sim: int = 10000, height_ratio: list = [1, 1.5], weight_factor: float = 1.0, **kwargs) → None