InlineExplainer

As a data scientist you often work inside a notebook environment where you quickly interactively like to explore your data. The InlineExplainer allows you to do this by running ExplainerComponents (or whole tabs) inline inside your Jupyter notebook (also works in Google Colab!).

_images/inline_screenshot.gif

This allows you to quickly check model performance, look for shap importances, etc. The components are sorted into subcategories and work with tab-completion.

Example use:

from explainerdashboard import InlineExplainer
ie = InlineExplainer(explainer)
ie.importances()
ie.model_stats()
ie.prediction()
ie.random_index()
ie.tab.importances()
ie.tab.modelsummary()
ie.tab.contributions()
ie.tab.dependence()
ie.tab.interactions()
ie.tab.decisiontrees()
ie.shap.overview()
ie.shap.summary()
ie.shap.dependence()
ie.shap.interaction_overview()
ie.shap.interaction_summary()
ie.shap.interaction_dependence()
ie.shap.contributions_graph()
ie.shap.contributions_table()
ie.classifier.model_stats()
ie.classifier.precision()
ie.classifier.confusion_matrix()
ie.classifier.lift_curve()
ie.classifier.classification()
ie.classifier.roc_auc()
ie.classifier.pr_auc()
ie.regression.model_stats()
ie.regression.pred_vs_actual()
ie.regression.residuals()
ie.regression.plots_vs_col()
ie.decisiontrees.overview()
ie.decisiontrees.decision_trees()
ie.decisiontrees.decisionpath_table()
ie.decisiontrees.decisionpath_graph()

You can also add options for the size of the output width, or to display the component in a separate tab (‘external’), or running on a different port:

InlineExplainer(explainer, mode='external', port=8051, width=1000, height=800).importances()

Note

You can run a component without instantiating the InlineExplainer first, like for example InlineExplainer(explainer).importances(), but then you cannot inspect the kwargs and docstring of that particular component. So to inspect kwargs and docstring you would run:

ie = InlineExplainer(explainer)
?ie.importances

(or alternatively hit shift-tab in jupyter of course)

You can kill an InlineExplainer running on a particular port with ExplainerDashboard.terminate(port=8050).

InlineExplainer documentation

class explainerdashboard.dashboards.InlineExplainer(explainer, mode='inline', width=1000, height=800, port=8050, **kwargs)

Run a single tab inline in a Jupyter notebook using specific method calls.

Parameters
  • explainer – an Explainer object

  • mode (str, optional) – either ‘inline’, ‘jupyterlab’ or ‘external’

  • width – width in pixels of inline iframe

  • height – height in pixels of inline iframe

  • port – port to run if mode=’external’

tab

subclass with InlineExplainerTabs layouts, e.g. InlineExplainer(explainer).tab.modelsummary()

shap

subclass with InlineShapExplainer layouts, e.g. InlineExplainer(explainer).shap.dependence()

classifier

subclass with InlineClassifierExplainer plots, e.g. InlineExplainer(explainer).classifier.confusion_matrix()

regression

subclass with InlineRegressionExplainer plots, e.g. InlineExplainer(explainer).regression.residuals()

decisiontrees

subclass with InlineDecisionTreesExplainer plots, e.g. InlineExplainer(explainer).decisiontrees.decisiontrees()

importances(title='Importances', name=None, subtitle='Which features had the biggest impact?', hide_type=False, hide_depth=False, hide_popout=False, hide_title=False, hide_subtitle=False, hide_selector=False, pos_label=None, importance_type='shap', depth=None, no_permutations=False, description=None)

Display features importances component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Feature Importances”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str, optional) – Subtitle.

  • hide_type (bool, optional) – Hide permutation/shap selector toggle. Defaults to False.

  • hide_depth (bool, optional) – Hide number of features toggle. Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • hide_title (bool, optional) – hide title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • importance_type (str, {'permutation', 'shap'} optional) – initial importance type to display. Defaults to “shap”.

  • depth (int, optional) – Initial number of top features to display. Defaults to None (=show all).

  • no_permutations (bool, optional) – Do not use the permutation importances for this component. Defaults to False.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

model_stats(title='Models Stats', **kwargs)

Runs model_stats inline in notebook

prediction(title='Prediction', name=None, hide_index=False, hide_percentile=False, hide_title=False, hide_subtitle=False, hide_selector=False, pos_label=None, index=None, percentile=True, description=None)

Shows a summary for a particular prediction

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Prediction Summary”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_index (bool, optional) – hide index selector. Defaults to False.

  • hide_percentile (bool, optional) – hide percentile toggle. Defaults to False.

  • hide_title (bool, optional) – hide title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ({int, str}, optional) – Index to display prediction summary for. Defaults to None.

  • percentile (bool, optional) – Whether to add the prediction percentile. Defaults to True.

random_index(title='Random Index', **kwargs)

show random index selector inline in notebook

pdp(title='Partial Dependence Plots', name=None, subtitle='How does the prediction change if you change one feature?', hide_col=False, hide_index=False, hide_title=False, hide_subtitle=False, hide_footer=False, hide_selector=False, hide_popout=False, hide_dropna=False, hide_sample=False, hide_gridlines=False, hide_gridpoints=False, hide_cats_sort=False, index_dropdown=True, feature_input_component=None, pos_label=None, col=None, index=None, dropna=True, sample=100, gridlines=50, gridpoints=10, cats_sort='freq', description=None)

Show Partial Dependence Plot component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Partial Dependence Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_col (bool, optional) – Hide feature selector. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_title (bool, optional) – Hide title, Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • hide_dropna (bool, optional) – Hide drop na’s toggle Defaults to False.

  • hide_sample (bool, optional) – Hide sample size input. Defaults to False.

  • hide_gridlines (bool, optional) – Hide gridlines input. Defaults to False.

  • hide_gridpoints (bool, optional) – Hide gridpounts input. Defaults to False.

  • hide_cats_sort (bool, optional) – Hide the categorical sorting dropdown. Defaults to False.

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • feature_input_component (FeatureInputComponent) – A FeatureInputComponent that will give the input to the graph instead of the index selector. If not None, hide_index=True. Defaults to None.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • col (str, optional) – Feature to display PDP for. Defaults to None.

  • index ({int, str}, optional) – Index to add ice line to plot. Defaults to None.

  • dropna (bool, optional) – Drop rows where values equal explainer.na_fill (usually -999). Defaults to True.

  • sample (int, optional) – Sample size to calculate average partial dependence. Defaults to 100.

  • gridlines (int, optional) – Number of ice lines to display in plot. Defaults to 50.

  • gridpoints (int, optional) – Number of breakpoints on horizontal axis Defaults to 10.

  • cats_sort (str, optional) – how to sort categories: ‘alphabet’, ‘freq’ or ‘shap’. Defaults to ‘freq’.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

class explainerdashboard.dashboards.InlineExplainerTabs(inline_explainer, name)
importances(title='Importances', name=None, hide_type=False, hide_depth=False, hide_cats=False, hide_title=False, hide_selector=False, pos_label=None, importance_type='shap', depth=None, cats=True, disable_permutations=False)

Overview tab of feature importances

Can show both permutation importances and mean absolute shap values.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Feature Importances”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_selector (bool, optional) –

  • importance_type (str, {'permutation', 'shap'} optional) – Type of importance to describe. Defaults to “shap”.

  • depth (int, optional) – Number of features to display by default. Defaults to None.

  • cats (bool, optional) – Group categoricals together. Defaults to True.

modelsummary(title='Model Summary', name=None, bin_size=0.1, quantiles=10, cutoff=0.5, logs=False, pred_or_actual='vs_pred', residuals='difference', col=None)

Tab shows a summary of model performance.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Model Performance”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • bin_size (float, optional) – precision plot bin_size. Defaults to 0.1.

  • quantiles (int, optional) – precision plot number of quantiles. Defaults to 10.

  • cutoff (float, optional) – cutoff for classifier plots. Defaults to 0.5.

  • logs (bool, optional) – use logs for regression plots. Defaults to False.

  • pred_or_actual (str, optional) – show residuals vs prediction or vs actual. Defaults to “vs_pred”.

  • residuals (str, {'difference', 'ratio', 'log-ratio'} optional) – How to calcualte residuals. Defaults to ‘difference’.

  • col ([type], optional) – Feature to show residuals against. Defaults to None.

contributions(title='Contributions', name=None)

Tab showing individual predictions, the SHAP contributions that add up to this predictions, in both graph and table form, and a pdp plot.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Individual Predictions”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • higher_is_better (bool, optional) – in contributions plot, up is green and down is red. (set to False to flip)

whatif(title='What if...', name=None)
Tab showing individual predictions and allowing edits

to the features…

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Individual Predictions”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

dependence(title='Shap Dependence', name=None, tab_id='shap_dependence', depth=None, cats=True)

Tab showing both a summary of feature importance (aggregate or detailed). for each feature, and a shap dependence graph.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to ‘Feature Dependence’.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

interactions(title='Shap Interactions', name=None, depth=None, cats=True)

[summary]

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to ‘Feature Interactions’.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • depth (int, optional) – default number of feature to display. Defaults to None.

  • cats (bool, optional) – default grouping of cats. Defaults to True.

decisiontrees(title='Decision Trees', name=None)

Tab showing individual decision trees

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to ‘Decision Trees’.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

class explainerdashboard.dashboards.InlineShapExplainer(inline_explainer, name)
overview(title='Shap Overview', name=None, hide_selector=True, hide_shapsummary=False, hide_shapdependence=False, depth=None)

Composite of ShapSummary and ShapDependence component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Feature Dependence”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_selector (bool, optional) – hide all pos label selectors. Defaults to True.

  • hide_shapsummary (bool, optional) – hide ShapSummaryComponent

  • hide_shapdependence (bool, optional) – ShapDependenceComponent

  • depth (int, optional) – Number of features to display. Defaults to None.

summary(title='Shap Summary', name=None, subtitle='Ordering features by shap value', hide_title=False, hide_subtitle=False, hide_depth=False, hide_type=False, hide_index=False, hide_selector=False, hide_popout=False, pos_label=None, depth=None, summary_type='aggregate', max_cat_colors=5, index=None, plot_sample=None, description=None)

Shows shap summary component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Shap Dependence Summary”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide the title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_depth (bool, optional) – hide the depth toggle. Defaults to False.

  • hide_type (bool, optional) – hide the summary type toggle (aggregated, detailed). Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • depth (int, optional) – initial number of features to show. Defaults to None.

  • summary_type (str, {'aggregate', 'detailed'}. optional) – type of summary graph to show. Defaults to “aggregate”.

  • max_cat_colors (int, optional) – for categorical features, maximum number of categories to label with own color. Defaults to 5.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

dependence(title='Shap Dependence', name=None, subtitle='Relationship between feature value and SHAP value', hide_title=False, hide_subtitle=False, hide_col=False, hide_color_col=False, hide_index=False, hide_selector=False, hide_outliers=False, hide_cats_topx=False, hide_cats_sort=False, hide_popout=False, hide_footer=False, pos_label=None, col=None, color_col=None, index=None, remove_outliers=False, cats_topx=10, cats_sort='freq', max_cat_colors=5, plot_sample=None, description=None)

Show shap dependence graph

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Shap Dependence”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide component title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_col (bool, optional) – hide feature selector. Defaults to False.

  • hide_color_col (bool, optional) – hide color feature selector Defaults to False.

  • hide_index (bool, optional) – hide index selector Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_cats_topx (bool, optional) – hide the categories topx input. Defaults to False.

  • hide_cats_sort (bool, optional) – hide the categories sort selector.Defaults to False.

  • hide_outliers (bool, optional) – Hide remove outliers toggle input. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • hide_footer (bool, optional) – hide the footer.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • col (str, optional) – Feature to display. Defaults to None.

  • color_col (str, optional) – Color plot by values of this Feature. Defaults to None.

  • index (int, optional) – Highlight a particular index. Defaults to None.

  • remove_outliers (bool, optional) – remove outliers in feature and color feature from the plot.

  • cats_topx (int, optional) – maximum number of categories to display for categorical features. Defaults to 10.

  • cats_sort (str, optional) – how to sort categories: ‘alphabet’, ‘freq’ or ‘shap’. Defaults to ‘freq’.

  • max_cat_colors (int, optional) – for categorical features, maximum number of categories to label with own color. Defaults to 5.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

interaction_overview(title='Interactions Overview', name=None, hide_selector=True, hide_interactionsummary=False, hide_interactiondependence=False, depth=None)

Composite of InteractionSummaryComponent and InteractionDependenceComponent

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Feature Interactions”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_selector (bool, optional) – hide all pos label selectors. Defaults to True.

  • hide_interactionsummary (bool, optional) – hide InteractionSummaryComponent

  • hide_interactiondependence (bool, optional) – hide InteractionDependenceComponent

  • depth (int, optional) – Initial number of features to display. Defaults to None.

interaction_summary(title='Shap Interaction Summary', name=None, subtitle='Ordering features by shap interaction value', hide_title=False, hide_subtitle=False, hide_col=False, hide_depth=False, hide_type=False, hide_index=False, hide_popout=False, hide_selector=False, pos_label=None, col=None, depth=None, summary_type='aggregate', max_cat_colors=5, index=None, plot_sample=None, description=None)

Show SHAP Interaciton values summary component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Interactions Summary”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide the component title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_col (bool, optional) – Hide the feature selector. Defaults to False.

  • hide_depth (bool, optional) – Hide depth toggle. Defaults to False.

  • hide_type (bool, optional) – Hide summary type toggle. Defaults to False.

  • hide_index (bool, optional) – Hide the index selector. Defaults to False

  • hide_popout (bool, optional) – hide popout button

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • col (str, optional) – Feature to show interaction summary for. Defaults to None.

  • depth (int, optional) – Number of interaction features to display. Defaults to None.

  • summary_type (str, {'aggregate', 'detailed'}, optional) – type of summary graph to display. Defaults to “aggregate”.

  • max_cat_colors (int, optional) – for categorical features, maximum number of categories to label with own color. Defaults to 5.

  • index (str) – Default index. Defaults to None.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

interaction_dependence(title='Shap Interaction Dependence', name=None, subtitle='Relation between feature value and shap interaction value', hide_title=False, hide_subtitle=False, hide_col=False, hide_interact_col=False, hide_index=False, hide_popout=False, hide_selector=False, hide_outliers=False, hide_cats_topx=False, hide_cats_sort=False, hide_top=False, hide_bottom=False, pos_label=None, col=None, interact_col=None, remove_outliers=False, cats_topx=10, cats_sort='freq', max_cat_colors=5, plot_sample=None, description=None, index=None)

Interaction Dependence Component.

Shows two graphs:

top graph: col vs interact_col bottom graph: interact_col vs col

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Interactions Dependence”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – Hide component title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_col (bool, optional) – Hide feature selector. Defaults to False.

  • hide_interact_col (bool, optional) – Hide interaction feature selector. Defaults to False.

  • hide_highlight (bool, optional) – Hide highlight index selector. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_outliers (bool, optional) – Hide remove outliers toggle input. Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • hide_cats_topx (bool, optional) – hide the categories topx input. Defaults to False.

  • hide_cats_sort (bool, optional) – hide the categories sort selector. Defaults to False.

  • hide_top (bool, optional) – Hide the top interaction graph (col vs interact_col). Defaults to False.

  • hide_bottom (bool, optional) – hide the bottom interaction graph (interact_col vs col). Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • col (str, optional) – Feature to find interactions for. Defaults to None.

  • interact_col (str, optional) – Feature to interact with. Defaults to None.

  • highlight (int, optional) – Index row to highlight Defaults to None.

  • remove_outliers (bool, optional) – remove outliers in feature and color feature from the plot.

  • cats_topx (int, optional) – number of categories to display for categorical features.

  • cats_sort (str, optional) – how to sort categories: ‘alphabet’, ‘freq’ or ‘shap’. Defaults to ‘freq’.

  • max_cat_colors (int, optional) – for categorical features, maximum number of categories to label with own color. Defaults to 5.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

contributions_graph(title='Contributions', name=None, subtitle='How has each feature contributed to the prediction?', hide_title=False, hide_subtitle=False, hide_index=False, hide_depth=False, hide_sort=False, hide_orientation=True, hide_selector=False, hide_popout=False, feature_input_component=None, index_dropdown=True, pos_label=None, index=None, depth=None, sort='high-to-low', orientation='vertical', higher_is_better=True, description=None)

Display Shap contributions to prediction graph component

Parameters
  • explainer (Explainer) – explainer object constructed , with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Contributions Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – Hide component title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_depth (bool, optional) – Hide depth toggle. Defaults to False.

  • hide_sort (bool, optional) – Hide the sorting dropdown. Defaults to False.

  • hide_orientation (bool, optional) – Hide the orientation dropdown. Defaults to True.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • feature_input_component (FeatureInputComponent) – A FeatureInputComponent that will give the input to the graph instead of the index selector. If not None, hide_index=True. Defaults to None.

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ({int, bool}, optional) – Initial index to display. Defaults to None.

  • depth (int, optional) – Initial number of features to display. Defaults to None.

  • sort ({'abs', 'high-to-low', 'low-to-high', 'importance'}, optional) – sorting of shap values. Defaults to ‘high-to-low’.

  • orientation ({'vertical', 'horizontal'}, optional) – orientation of bar chart. Defaults to ‘vertical’.

  • higher_is_better (bool, optional) – Color positive shap values green and negative shap values red, or the reverse.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

contributions_table(title='Contributions', name=None, subtitle='How has each feature contributed to the prediction?', hide_title=False, hide_subtitle=False, hide_index=False, hide_depth=False, hide_sort=False, hide_selector=False, feature_input_component=None, index_dropdown=True, pos_label=None, index=None, depth=None, sort='abs', description=None)

Show SHAP values contributions to prediction in a table component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Contributions Table”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – Hide component title. Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_depth (bool, optional) – Hide depth selector. Defaults to False.

  • hide_sort (bool, optional) – Hide sorting dropdown. Default to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • feature_input_component (FeatureInputComponent) – A FeatureInputComponent that will give the input to the graph instead of the index selector. If not None, hide_index=True. Defaults to None.

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ([type], optional) – Initial index to display. Defaults to None.

  • depth ([type], optional) – Initial number of features to display. Defaults to None.

  • sort ({'abs', 'high-to-low', 'low-to-high', 'importance'}, optional) – sorting of shap values. Defaults to ‘high-to-low’.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

class explainerdashboard.dashboards.InlineClassifierExplainer(inline_explainer, name)
model_stats(title='Models Stats', name=None, hide_title=True, hide_selector=True, hide_globalcutoff=False, hide_modelsummary=False, hide_confusionmatrix=False, hide_precision=False, hide_classification=False, hide_rocauc=False, hide_prauc=False, hide_liftcurve=False, hide_cumprecision=False, pos_label=None, bin_size=0.1, quantiles=10, cutoff=0.5)
Composite of multiple classifier related components:
  • precision graph

  • confusion matrix

  • lift curve

  • classification graph

  • roc auc graph

  • pr auc graph

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Decision Trees”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_title (bool, optional) – hide title. Defaults to True.

  • hide_selector (bool, optional) – hide all pos label selectors. Defaults to True.

  • hide_globalcutoff (bool, optional) – hide CutoffPercentileComponent

  • hide_modelsummary (bool, optional) – hide ClassifierModelSummaryComponent

  • hide_confusionmatrix (bool, optional) – hide ConfusionMatrixComponent

  • hide_precision (bool, optional) – hide PrecisionComponent

  • hide_classification (bool, optional) – hide ClassificationComponent

  • hide_rocauc (bool, optional) – hide RocAucComponent

  • hide_prauc (bool, optional) – hide PrAucComponent

  • hide_liftcurve (bool, optional) – hide LiftCurveComponent

  • hide_cumprecision (bool, optional) – hide CumulativePrecisionComponent

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • bin_size (float, optional) – bin_size for precision plot. Defaults to 0.1.

  • quantiles (int, optional) – number of quantiles for precision plot. Defaults to 10.

  • cutoff (float, optional) – initial cutoff. Defaults to 0.5.

precision(title='Precision Plot', name=None, subtitle='Does fraction positive increase with predicted probability?', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_binsize=False, hide_binmethod=False, hide_multiclass=False, hide_selector=False, hide_popout=False, pos_label=None, bin_size=0.1, quantiles=10, cutoff=0.5, quantiles_or_binsize='bin_size', multiclass=False, description=None)

Shows a precision graph with toggles.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Precision Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – Hide cutoff slider. Defaults to False.

  • hide_binsize (bool, optional) – hide binsize/quantiles slider. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_binmethod (bool, optional) – Hide binsize/quantiles toggle. Defaults to False.

  • hide_multiclass (bool, optional) – Hide multiclass toggle. Defaults to False.

  • hide_selector – Hide pos label selector. Default to True.

  • hide_popout (bool, optional) – hide popout button

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • bin_size (float, optional) – Size of bins in probability space. Defaults to 0.1.

  • quantiles (int, optional) – Number of quantiles to divide plot. Defaults to 10.

  • cutoff (float, optional) – Cutoff to display in graph. Defaults to 0.5.

  • quantiles_or_binsize (str, {'quantiles', 'bin_size'}, optional) – Default bin method. Defaults to ‘bin_size’.

  • multiclass (bool, optional) – Display all classes. Defaults to False.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

cumulative_precision(title='Cumulative Precision Plot', name=None, subtitle='Expected distribution for highest scores', hide_title=False, hide_subtitle=False, hide_footer=False, hide_selector=False, hide_popout=False, pos_label=None, hide_cutoff=False, cutoff=None, hide_percentile=False, percentile=None, description=None)

Show cumulative precision component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Cumulative Precision”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide the title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

confusion_matrix(title='Confusion Matrix', name=None, subtitle='How many false positives and false negatives?', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_percentage=False, hide_binary=False, hide_selector=False, hide_popout=False, hide_normalize=False, normalize='all', pos_label=None, cutoff=0.5, percentage=True, binary=True, description=None)

Display confusion matrix component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Confusion Matrix”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – Hide cutoff slider. Defaults to False.

  • hide_percentage (bool, optional) – Hide percentage toggle. Defaults to False.

  • hide_binary (bool, optional) – Hide binary toggle. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • cutoff (float, optional) – Default cutoff. Defaults to 0.5.

  • percentage (bool, optional) – Display percentages instead of counts. Defaults to True.

  • binary (bool, optional) – Show binary instead of multiclass confusion matrix. Defaults to True.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

  • normalize (str[‘true’, ‘pred’, ‘all’]) – normalizes confusion matrix over the true (rows), predicted (columns) conditions or all the population. Defaults to all

lift_curve(title='Lift Curve', name=None, subtitle='Performance how much better than random?', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_percentage=False, hide_wizard=False, hide_selector=False, hide_popout=False, pos_label=None, cutoff=0.5, percentage=True, wizard=True, description=None)

Show liftcurve component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Lift Curve”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – Hide cutoff slider. Defaults to False.

  • hide_percentage (bool, optional) – Hide percentage toggle. Defaults to False.

  • hide_wizard (bool, optional) – hide the wizard toggle. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • cutoff (float, optional) – Cutoff for lift curve. Defaults to 0.5.

  • percentage (bool, optional) – Display percentages instead of counts. Defaults to True.

  • wizard (bool, optional) – display the wizard in the graph.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

classification(title='Classification', name=None, subtitle='Distribution of labels above and below cutoff', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_percentage=False, hide_selector=False, hide_popout=False, pos_label=None, cutoff=0.5, percentage=True, description=None)

Shows a barchart of the number of classes above the cutoff and below the cutoff.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Classification Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide the title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – Hide cutoff slider. Defaults to False.

  • hide_percentage (bool, optional) – Hide percentage toggle. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • cutoff (float, optional) – Cutoff for prediction. Defaults to 0.5.

  • percentage (bool, optional) – Show percentage instead of counts. Defaults to True.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

roc_auc(title='ROC AUC Curve', name=None, subtitle='Trade-off between False positives and false negatives', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_selector=False, hide_popout=False, pos_label=None, cutoff=0.5, description=None)

Show ROC AUC curve component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “ROC AUC Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – Hide cutoff slider. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • cutoff (float, optional) – default cutoff. Defaults to 0.5.

pr_auc(title='PR AUC Curve', name=None, subtitle='Trade-off between Precision and Recall', hide_title=False, hide_subtitle=False, hide_footer=False, hide_cutoff=False, hide_selector=False, hide_popout=False, pos_label=None, cutoff=0.5, description=None)

Display PR AUC plot component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “PR AUC Plot”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_cutoff (bool, optional) – hide cutoff slider. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selector. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • cutoff (float, optional) – default cutoff. Defaults to 0.5.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

class explainerdashboard.dashboards.InlineRegressionExplainer(inline_explainer, name)
model_stats(title='Models Stats', name=None, hide_title=True, hide_modelsummary=False, hide_predsvsactual=False, hide_residuals=False, hide_regvscol=False, logs=False, pred_or_actual='vs_pred', residuals='difference', col=None)

Composite for displaying multiple regression related graphs:

  • predictions vs actual plot

  • residual plot

  • residuals vs feature

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Regression Stats”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_title (bool, optional) – hide title. Defaults to True.

  • hide_modelsummary (bool, optional) – hide RegressionModelSummaryComponent

  • hide_predsvsactual (bool, optional) – hide PredictedVsActualComponent

  • hide_residuals (bool, optional) – hide ResidualsComponent

  • hide_regvscol (bool, optional) – hide RegressionVsColComponent

  • logs (bool, optional) – Use log axis. Defaults to False.

  • pred_or_actual (str, optional) – plot residuals vs predictions or vs y (actual). Defaults to “vs_pred”.

  • residuals (str, {'difference', 'ratio', 'log-ratio'} optional) – How to calcualte residuals. Defaults to ‘difference’.

  • col ({str, int}, optional) – Feature to use for residuals plot. Defaults to None.

pred_vs_actual(title='Predicted vs Actual', name=None, subtitle='How close is the predicted value to the observed?', hide_title=False, hide_subtitle=False, hide_log_x=False, hide_log_y=False, hide_popout=False, logs=False, log_x=False, log_y=False, round=3, plot_sample=None, description=None)

Shows a plot of predictions vs y.

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Predicted vs Actual”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) –

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_log_x (bool, optional) – Hide the log_x toggle. Defaults to False.

  • hide_log_y (bool, optional) – Hide the log_y toggle. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • logs (bool, optional) – Whether to use log axis. Defaults to False.

  • log_x (bool, optional) – log only x axis. Defaults to False.

  • log_y (bool, optional) – log only y axis. Defaults to False.

  • round (int, optional) – rounding to apply to float predictions. Defaults to 3.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

residuals(title='Residuals', name=None, subtitle='How much is the model off?', hide_title=False, hide_subtitle=False, hide_footer=False, hide_pred_or_actual=False, hide_ratio=False, hide_popout=False, pred_or_actual='vs_pred', residuals='difference', round=3, plot_sample=None, description=None)

Residuals plot component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Residuals”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) –

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_pred_or_actual (bool, optional) – hide vs predictions or vs actual for x-axis toggle. Defaults to False.

  • hide_ratio (bool, optional) – hide residual type dropdown. Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • pred_or_actual (str, {'vs_actual', 'vs_pred'}, optional) – Whether to plot actual or predictions on the x-axis. Defaults to “vs_pred”.

  • residuals (str, {'difference', 'ratio', 'log-ratio'} optional) – How to calcualte residuals. Defaults to ‘difference’.

  • round (int, optional) – rounding to apply to float predictions. Defaults to 3.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

plots_vs_col(title='Plots vs col', name=None, subtitle='Are predictions and residuals correlated with features?', hide_title=False, hide_subtitle=False, hide_footer=False, hide_col=False, hide_ratio=False, hide_points=False, hide_winsor=False, hide_cats_topx=False, hide_cats_sort=False, hide_popout=False, col=None, display='difference', round=3, points=True, winsor=0, cats_topx=10, cats_sort='freq', plot_sample=None, description=None)

Show residuals, observed or preds vs a particular Feature component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Plot vs feature”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) –

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_footer (bool, optional) – hide the footer at the bottom of the component

  • hide_col (bool, optional) – Hide de column selector. Defaults to False.

  • hide_ratio (bool, optional) – Hide the toggle. Defaults to False.

  • hide_points (bool, optional) – Hide group points toggle. Defaults to False.

  • hide_winsor (bool, optional) – Hide winsor input. Defaults to False.

  • hide_cats_topx (bool, optional) – hide the categories topx input. Defaults to False.

  • hide_cats_sort (bool, optional) – hide the categories sort selector.Defaults to False.

  • hide_popout (bool, optional) – hide popout button. Defaults to False.

  • col ([type], optional) – Initial feature to display. Defaults to None.

  • display (str, {'observed', 'predicted', difference', 'ratio', 'log-ratio'} optional) – What to display on y axis. Defaults to ‘difference’.

  • round (int, optional) – rounding to apply to float predictions. Defaults to 3.

  • points (bool, optional) – display point cloud next to violin plot for categorical cols. Defaults to True

  • winsor (int, 0-50, optional) – percentage of outliers to winsor out of the y-axis. Defaults to 0.

  • cats_topx (int, optional) – maximum number of categories to display for categorical features. Defaults to 10.

  • cats_sort (str, optional) – how to sort categories: ‘alphabet’, ‘freq’ or ‘shap’. Defaults to ‘freq’.

  • plot_sample (int, optional) – Instead of all points only plot a random sample of points. Defaults to None (=all points)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

class explainerdashboard.dashboards.InlineDecisionTreesExplainer(inline_explainer, name)
overview(title='Decision Trees', name=None, hide_treeindexselector=False, hide_treesgraph=False, hide_treepathtable=False, hide_treepathgraph=False, hide_selector=True, index_check=True)

Composite of decision tree related components:

  • index selector

  • individual decision trees barchart

  • decision path table

  • deciion path graph

Parameters
  • explainer (Explainer) – explainer object constructed with either RandomForestClassifierExplainer() or RandomForestRegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Decision Trees”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • hide_treeindexselector (bool, optional) – hide ClassifierRandomIndexComponent or RegressionRandomIndexComponent

  • hide_treesgraph (bool, optional) – hide DecisionTreesComponent

  • hide_treepathtable (bool, optional) – hide DecisionPathTableComponent

  • hide_treepathgraph (bool, optional) – DecisionPathGraphComponent

  • hide_selector (bool, optional) – hide all pos label selectors. Defaults to True.

  • index_check (bool, optional) – only pass valid indexes from random index selector to feature input. Defaults to True.

decisiontrees(title='Decision Trees', name=None, subtitle='Displaying individual decision trees', hide_title=False, hide_subtitle=False, hide_index=False, hide_highlight=False, hide_selector=False, hide_popout=False, index_dropdown=True, pos_label=None, index=None, highlight=None, higher_is_better=True, description=None)

Show prediction from individual decision trees inside RandomForest component

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Decision Trees”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title, Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_highlight (bool, optional) – Hide tree highlight selector. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • hide_popout (bool, optional) – hide popout button

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ({str, int}, optional) – Initial index to display. Defaults to None.

  • highlight (int, optional) – Initial tree to highlight. Defaults to None.

  • higher_is_better (bool, optional) – up is green, down is red. If False flip the colors. (for gbm models only)

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

decisionpath_table(title='Decision path', name=None, subtitle='Decision path through decision tree', hide_title=False, hide_subtitle=False, hide_index=False, hide_highlight=False, hide_selector=False, index_dropdown=True, pos_label=None, index=None, highlight=None, description=None)

Display a table of the decision path through a particular decision tree

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Decision path table”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title, Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_highlight (bool, optional) – Hide tree index selector. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ({str, int}, optional) – Initial index to display decision path for. Defaults to None.

  • highlight (int, optional) – Initial tree idx to display decision path for. Defaults to None.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.

decisionpath_graph(title='Decision path', name=None, subtitle='Decision path through decision tree', hide_title=False, hide_subtitle=False, hide_index=False, hide_highlight=False, hide_selector=False, index_dropdown=True, pos_label=None, index=None, highlight=None, description=None)

Display a table of the decision path through a particular decision tree

Parameters
  • explainer (Explainer) – explainer object constructed with either ClassifierExplainer() or RegressionExplainer()

  • title (str, optional) – Title of tab or page. Defaults to “Decision path table”.

  • name (str, optional) – unique name to add to Component elements. If None then random uuid is generated to make sure it’s unique. Defaults to None.

  • subtitle (str) – subtitle

  • hide_title (bool, optional) – hide title, Defaults to False.

  • hide_subtitle (bool, optional) – Hide subtitle. Defaults to False.

  • hide_index (bool, optional) – Hide index selector. Defaults to False.

  • hide_highlight (bool, optional) – Hide tree index selector. Defaults to False.

  • hide_selector (bool, optional) – hide pos label selectors. Defaults to False.

  • index_dropdown (bool, optional) – Use dropdown for index input instead of free text input. Defaults to True.

  • pos_label ({int, str}, optional) – initial pos label. Defaults to explainer.pos_label

  • index ({str, int}, optional) – Initial index to display decision path for. Defaults to None.

  • highlight (int, optional) – Initial tree idx to display decision path for. Defaults to None.

  • description (str, optional) – Tooltip to display when hover over component title. When None default text is shown.