seqgra.comparator.roccomparator module

Creates ROC curves of different models

Classes:

• ROCComparator: creates ROC curves from various grammars and architectures

class ROCComparator(analysis_id: str, output_dir: str, model_labels: Optional[List[str]] = None, silent: bool = False)

Bases: seqgra.comparator.comparator.Comparator

compare_models(grammar_ids: Optional[List[str]] = None, model_ids: Optional[List[str]] = None, set_names: Optional[List[str]] = None) → None

create_roc_curve(fpr: List[List[float]], tpr: List[List[float]], roc_auc: List[float], model_labels: List[str], file_name: str) → None

create_single_roc_curve(y_true, y_hat) → None

Create ROC curve.

Plots ROC curves for each class label, including micro-average and macro-average. Saves plot as PDF in file_name.

Parameters

• y_true (array) – TODO ; shape = [n_samples, n_classes]

• y_hat (array) – TODO ; shape = [n_samples, n_classes]

get_labels(grammar_id: str, model_id: str, set_name: str) → int