mergernet.visualization.plot.roc#

mergernet.visualization.plot.roc(true_series, pred_series, curves, zoom_range=None, zoom_pos=None, filename=None, show=True)[source]#

Plot ROC curve of a model :param true_series: The series of true labels.

Shape of each serie: (n, n_class), where rep is the number of repetitions (std curve computation), n is the number of examples and n_class is the number of classes.

Parameters:

  • pred_series (list of Serie) – The series of predictions. Shape of each serie: (n, n_class), where rep is the number of repetition (std curve computation), n is the number of examples and n_class is the number of classes.

  • curves (list) – Values can include: 0 … n_class, “macro” or “micro”. If curve is a int, then a curve of the specific class will be ploted. If curve is “macro”, the macro-average of all classes will be computed. If curve is “micro”, the micro-average of all classes will be computed.

Notes

assert len(true_series) == len(pred_series) == len(curves)