Plotting Module

The module includes basic plotting functionality for machine learning applications. Namely, following functions are implemented:

scripts.plotting.plot_1d_decision_regions(x, y, model, meshsize=0.01, title='unnamed', ax=None)

Plots 1d decision regions for given feature and model. In addition, it adds on top training datapoints.

Parameters
  • x (numpy.ndarray) – One dimensional array with values of the given feature.

  • y (numpy.ndarray) – One dimensional array with labels of given classes from training dataset.

  • model (python object) – Trained model.

  • meshsize (float, optional) – Space between points wihthin the meshgrid.

  • title (str, optional) – Title of the plot.

  • ax (matplotlib.axes.Axes, optional) – If you want to plot it to the prespecified axes.

Returns

fig – The plotted 1d decision region.

Return type

matplotlib.figure.Figure

Raises

AssertionError – If the feature space is not one dimensional.

scripts.plotting.plot_2d_decision_regions(x, y, model, meshsize=100, marker='s', colors=None, show_probs=False, title='unnamed', ax=None)

Plots 2d decision regions for given 2 features and model. In addition, it adds on top training datapoints.

Parameters
  • x (numpy.ndarray) – 2 dimensional array with values of the given features.

  • y (numpy.ndarray) – One dimensional array with labels of given classes from training dataset.

  • model (python object) – Trained model.

  • meshsize (float, optional) – Space between points wihthin the meshgrid.

  • marker (str, optional) – Type of the marker to represent datapoints in the plot.

  • colors (list, optional) – List with colors for each feature.

  • show_probs (bool) – Intensity of the regions color is based on the probability of the predicted class.

  • title (str, optional) – Title of the plot.

  • ax (matplotlib.axes.Axes, optional) – If you want to plot it to the prespecified axes.

Returns

ax – The plotted 2d decision region.

Return type

matplotlib.axes.Axes

Raises

AssertionError – If the feature space is not two dimensional.