Seismometer

Exploration Widgets

ExploreSubgroups(**kwargs)

Explore the models base statistics based on a selected subpopulation.

ExploreModelEvaluation(**kwargs)

Exploration widget for model evaluation, showing model performance for a specific subpopulation.

ExploreCohortEvaluation(**kwargs)

Exploration widget for cohort evaluation, showing model performance across thresholds and cohort subgroups.

ExploreCohortHistograms(**kwargs)

Exploration widget to show the true positives and negative by model score.

ExploreCohortLeadTime(**kwargs)

Exploration widget for the lead time between a model prediction and an event of interest.

ExploreCohortOutcomeInterventionTimes(**kwargs)

Exploration widget for viewing rates of interventions and outcomes across categories in a cohort group.

ExploreFairnessAudit(**kwargs)

Exploration widget for model fairness across cohorts for a binary classifier.

ExploreModelScoreComparison(**kwargs)

Exploration widget for model evaluation, showing model performance for a specific subpopulation.

ExploreModelTargetComparison(**kwargs)

Exploration widget for model target evaluation, showing model performance for a specific subpopulation.

ExploreBinaryModelMetrics(**kwargs)

Explore the models performance metrics based on a selected metric.

Public APIs

cohort_comparison_report([exclude_cols])

Generates (or loads from disk) the ydata-profiling report stratified by the selected cohort variable.

cohort_list()

Displays an exhaustive list of available cohorts for analysis.

feature_alerts([exclude_cols])

Generates (or loads from disk) the ydata-profiling report feature quality alerts.

feature_summary([exclude_cols, inline])

Generates (or loads from disk) the ydata-profiling report.

model_evaluation([per_context_id])

Displays overall performance of the model.

plot_cohort_evaluation(cohort_col, ...[, ...])

Plots model performance metrics split by on a cohort attribute.

plot_cohort_hist()

Display a histogram plot of predicted probabilities for all cohorts in the selected attribute.

plot_leadtime_enc([score, ref_time, ...])

Displays the amount of time that a high prediction gives before an event of interest.

plot_binary_classifier_metrics(...[, ...])

Generates a plot with model metrics.

plot_trend_intervention_outcome()

Plots two timeseries based on selectors; an outcome and then an intervention.

show_info([plot_help])

Displays information about the dataset

show_cohort_summaries([by_target, by_score])

Displays a table of selectable attributes and their associated counts.

target_feature_summary([exclude_cols, inline])

Generates (or loads from disk) the ydata-profiling report stratified by the target variable.

Custom Visualization Controls

explore.ExplorationWidget(**kwargs)

Parent class for model exploration widgets.

explore.ExplorationSubpopulationWidget(**kwargs)

A widget for exploring subpopulations based on cohort selection.

explore.ExplorationModelSubgroupEvaluationWidget(...)

A widget for exploring the model performance of a cohort.

explore.ExplorationCohortSubclassEvaluationWidget(...)

A widget for exploring the model performance based on the subgroups of a cohort column.

explore.ExplorationCohortOutcomeInterventionEvaluationWidget(...)

A widget for exploring the model outcomes and interventions based on the subgroups of a cohort column.

explore.ExplorationScoreComparisonByCohortWidget(...)

A widget to explore different model scores based on a cohort selection.

explore.ExplorationTargetComparisonByCohortWidget(...)

A widget to explore different model targets based on a cohort selection.

explore.ExplorationMetricWidget(**kwargs)

A widget to explore different model metrics for a score, target, and cohort

Notebook and Configuration Helper APIs

download_example_dataset(dataset_name[, ...])

Downloads an example dataset from the specified branch to local data/ directory.

generate_dictionary_from_parquet(inpath, outpath)

Generate a data dictionary YAML file from a parquet file.