API Reference

Classification Plots

Plots to evaluate classification models.

Module containing all classification model evaluation plots

mlplot.classification.calibration(y_true, y_pred, ax=None, n_bins='auto')

Plot a calibration plot

Calibration plots are used the determine how well the predicted values match the true value.

This plot is as found in sklean.

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)
  • n_bins (int or string) – The number of bins to group y_pred. See numpy.histogram
mlplot.classification.confusion_matrix(y_true, y_pred, class_labels=None, threshold=0.5, ax=None)

Plot a heatmap for the confusion matrix

An example of this heatmap can be found on sklean.

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)
  • threshold (float) – Defines the cutoff to be considered in the asserted class
mlplot.classification.population_histogram(y_true, y_pred, class_labels=None, ax=None)

Plot histograms of the predictions grouped by class

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)
mlplot.classification.precision_recall(y_true, y_pred, x_axis='recall', ax=None)

Plot the precision-recall curve

An example of this plot can be found on sklean.

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)
  • x_axis (str ‘recall’ or ‘threshold’) – Specify the x axis of the plot. Precision recall tends to come in 2 flavors, one precision vs recall and the other precion and recall vs threshold.
mlplot.classification.report_table(y_true, y_pred, class_labels=None, ax=None)

Generate a report table containing key stats about the dataset

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)
mlplot.classification.roc_curve(y_true, y_pred, ax=None)

Reciever operating curve

Parameters:
  • y_true (np.array of str or int) – A vector of size N that contains the true labels. There should be two labels of type string or numeric.
  • y_pred (np.array of float) – A vector of size N that contains predictions as floats from 0 to 1.
  • class_labels (dict, optional) – A dictionary mapping from lables in y_true to class names. Ex: {0: ‘not dog’, 1: ‘is dog’}
  • ax (matplotlib.axes.Axes, optional)