tabensemb.trainer.Trainer.plot_partial_dependence#

method

Trainer.plot_partial_dependence(program: str, model_name: str, feature: str, ax=None, refit: bool = True, log_trans: bool = True, lower_lim: float = 2, upper_lim: float = 7, n_bootstrap: int = 1, grid_size: int = 30, CI: float = 0.95, verbose: bool = True, figure_kwargs: Dict | None = None, plot_kwargs: Dict | None = None, fill_between_kwargs: Dict | None = None, bar_kwargs: Dict | None = None, hist_kwargs: Dict | None = None, savefig_kwargs: Dict | None = None, save_show_close: bool = True) Axes[source]#

Calculate and plot a partial dependence plot with bootstrapping for a feature.

Parameters:
program

The selected model base.

model_name

The selected model in the model base.

feature

The selected feature to calculate partial dependence.

ax

matplotlib.axes.Axes

refit

Whether to refit models on bootstrapped datasets. See _bootstrap_fit().

log_trans

Whether the label data is in log scale.

lower_lim

Lower limit of all pdp plots.

upper_lim

Upper limit of all pdp plot.

n_bootstrap

The number of bootstrap evaluations. It should be greater than 0.

grid_size

The number of steps of all pdp plot.

CI

The confidence interval of pdp results calculated across multiple bootstrap runs.

verbose

Verbosity

figure_kwargs

Arguments for plt.figure.

plot_kwargs

Arguments for ax.plot.

fill_between_kwargs

Arguments for ax.fill_between.

bar_kwargs

Arguments for ax.bar (used for frequencies of categorical features).

hist_kwargs

Arguments for ax.hist (used for histograms of continuous features).

savefig_kwargs

Arguments for plt.savefig

save_show_close

Whether to save, show (in the notebook), and close the figure if ax is not given.

Returns:
matplotlib.axes.Axes