familiarData-class: Dataset obtained after evaluating models on a dataset.

In familiar: End-to-End Automated Machine Learning and Model Evaluation

Dataset obtained after evaluating models on a dataset.

Description

A familiarData object is created by evaluating familiarEnsemble or familiarModel objects on a dataset. Multiple familiarData objects are aggregated in a familiarCollection object.

Slots

name

Name of the dataset, e.g. training or internal validation.

outcome_type

Outcome type of the data used to create the object.

outcome_info

Outcome information object, which contains additional information concerning the outcome, such as class levels.

fs_vimp

Variable importance data collected from feature selection methods.

model_vimp

Variable importance data collected from model-specific algorithms implemented by models created by familiar.

permutation_vimp

Data collected for permutation variable importance.

hyperparameters

Hyperparameters collected from created models.

hyperparameter_data

Additional data concerning hyperparameters. This is currently not used yet.

required_features

The set of features required for complete reproduction, i.e. with imputation.

model_features

The set of features that are required for using the model or ensemble of models, but without imputation.

learner

Learning algorithm used to create the model or ensemble of models.

fs_method

Feature selection method used to determine variable importance for the model or ensemble of models.

pooling_table

Run table for the data underlying the familiarData object. Used internally.

prediction_data

Model predictions for a model or ensemble of models for the underlying dataset.

confusion_matrix

Confusion matrix for a model or ensemble of models, based on the underlying dataset.

decision_curve_data

Decision curve analysis data for a model or ensemble of models, based on the underlying dataset.

calibration_info

Calibration information, e.g. baseline survival in the development cohort.

calibration_data

Calibration data for a model or ensemble of models, based on the underlying dataset.

model_performance

Model performance data for a model or ensemble of models, based on the underlying dataset.

km_info

Information concerning risk-stratification cut-off values..

km_data

Kaplan-Meier survival data for a model or ensemble of models, based on the underlying dataset.

auc_data

AUC-ROC and AUC-PR data for a model or ensemble of models, based on the underlying dataset.

ice_data

Individual conditional expectation data for features included in a model or ensemble of models, based on the underlying dataset. Partial dependence data are computed on the fly from these data.

univariate_analysis

Univariate analysis of the underlying dataset.

feature_expressions

Feature expression values of the underlying dataset.

feature_similarity

Feature similarity information of the underlying dataset.

sample_similarity

Sample similarity information of the underlying dataset.

is_validation

Signifies whether the underlying data forms a validation dataset. Used internally.

generating_ensemble

Name of the ensemble that was used to generate the familiarData object.

project_id

Identifier of the project that generated the familiarData object.

familiar_version

Version of the familiar package.

familiarData objects contain information obtained by evaluating a single model or single ensemble of models on a dataset.

familiar documentation built on Sept. 30, 2024, 9:18 a.m.