API Reference

Package Sections

Built-in Evaluators

Ready-to-use evaluator implementations for common ML tasks.

Built-in Evaluators

Command Line Interface

Command line interface for Brisk.

Command Line Interface

Configuration

Interface for defining what models to train.

Configuration

Data

Handles data splitting and preprocessing.

Data

Data Preprocessing

Applies transformations to a dataset before training.

Data Preprocessing

Evaluation

Classes providing methods to evaluate models and plot results.

Evaluation

Reporting

Generates an interactive report from training results.

Reporting

Services

Infrastructure services including logging, I/O, and metadata management.

Services

Theme

Styling and file format settings for plots.

Theme

Training

Brings together the data and configuration to train models.

Training

API Objects

Object	Description
AlgorithmCollection	A collection of AlgorithmWrappers.
AlgorithmWrapper	Wraps a machine learning algorithm and provides an interface using the algorithm.
BarPlot	Creates bar plots for categorical features.
BaseEvaluator	Abstract base class for all evaluators.
BaseService	Abstract base class for all services in the Brisk framework.
CaptureStrategy	Strategy for capturing experiment state for reruns.
CategoricalStatistics	Computes statistics for categorical variables in datasets.
CompareModels	Compares performance across multiple models.
Configuration	Provide an interface for creating experiment groups.
ConfigurationManager	Process the ExperimentGroups and prepare the required DataManagers.
ConfusionMatrix	Computes confusion matrix for classification models.
ContinuousStatistics	Computes statistics for continuous variables in datasets.
CoordinatingStrategy	Strategy for coordinating multiple rerun operations.
CorrelationMatrix	Creates correlation matrix plots.
DataManager	Handles data splitting and preprocessing pipelines. Arguments are used to define the splitting strategy and preprocessing steps.
DataSplitInfo	Stores and analyzes training and testing datasets, providing methods for calculating descriptive statistics and visualizing feature distributions.
DataSplits	Stores DataSplitInfo instances.
Dataset	Represents a dataset in the report.
DatasetMeasureEvaluator	Base class for evaluators that compute dataset-level measures.
DatasetPlotEvaluator	Base class for evaluators that create dataset-level plots.
EnvironmentDiff	Represents the differences between environments.
EnvironmentManager	Manages environment capture, comparison, and export for reproducible runs.
EvaluateModel	Evaluates model performance using specified metrics.
EvaluateModelCV	Evaluates model performance using cross-validation.
EvaluationManager	Provides methods for evaluating models and generating plots.
EvaluatorRegistry	Registry for managing and discovering evaluators.
Experiment	Stores all the data needed for one experiment run.
ExperimentFactory	Create a que of Experiments from an ExperimentGroup.
ExperimentGroup	Groups experiments that will be run with the same settings.
FeatureDistribution	Data structure for feature distribution information.
FileFormatter	Formats log messages with a visual separator between log entries.
GlobalServiceManager	Manages global service instances and dependencies.
Histogram	Creates histogram plots for dataset features.
HyperparameterTuning	Performs hyperparameter optimization.
IOService	Provides file input/output operations and data serialization.
LoggingService	Manages logging configuration and handlers.
MeasureEvaluator	Base class for evaluators that compute model performance measures.
MetadataService	Manages experiment metadata and versioning information.
MetricManager	Stores MetricWrapper instances that define evaluation metrics.
MetricWrapper	Wraps a metric function and provides a convenient interface using the metric.
Navbar	Navigation bar configuration for reports.
NumpyEncoder	JSON encoder for NumPy arrays and data types.
PackageInfo	Information about a package and its version.
PickleJSONDecoder	JSON decoder that handles pickled objects.
PickleJSONEncoder	JSON encoder that handles pickled objects.
PlotConfusionHeatmap	Creates confusion matrix heatmap plots.
PlotData	Represents plot data and metadata for reports.
PlotEvaluator	Base class for evaluators that create model performance plots.
PlotFeatureImportance	Creates feature importance plots.
PlotLearningCurve	Creates learning curve plots.
PlotModelComparison	Creates model comparison plots.
PlotPrecisionRecallCurve	Creates precision-recall curve plots.
PlotPredVsObs	Creates predicted vs observed plots for regression models.
PlotResiduals	Creates residual plots for regression models.
PlotRocCurve	Creates ROC curve plots for classification models.
PlotSettings	Configuration for plot appearance and styling.
PlotShapleyValues	Creates SHAP value plots for model interpretability.
ReportData	Container for all report data and metadata.
ReportRenderer	Renders HTML reports from training results.
ReportingContext	Context manager for report generation operations.
ReportingService	Coordinates report generation and data collection.
RerunService	Manages experiment rerun capabilities and strategies.
RerunStrategy	Abstract base class for rerun strategies.
RoundedModel	Base model with automatic number rounding for display.
ServiceBundle	Bundles related services together for easier management.
TableData	Represents tabular data for report display.
ThemePickleJSONSerializer	Serializes theme objects using pickle and JSON encoding.
TqdmLoggingHandler	Logs messages to stdout or stderr using tqdm.
TrainingManager	Coordinates the training process, loading the data and running the experiments.
UtilityService	Provides common utility functions and helpers.
VersionMatch	Enumeration of version matching states.
Workflow	Defines the steps to take when training a model.
check_env	Checks the environment compatibility with a previous run.
cli_helpers	Provides helper functions for the CLI.
create	Creates a new project.
create_data	Creates a synthetic dataset.
export_env	Create a requirements.txt file from the environment captured during a previous experiment run.
find_project_root	Finds the project root directory containing .briskconfig.
load_data	Load a scikit-learn dataset by name.
load_sklearn_dataset	Load a scikit-learn dataset by name.
run	Run the current experiment setup.