skika.evaluation.inspect_recurrence#

Classes

DataExposure(data)

Class for accessing information from data at each observation"

InspectPrequential([max_samples, ...])

Class to run classifier and pass information to the visualizer.

InspectorVisualizer([name])

Class to visualize and plot online classification results in real time.

ModelExposure(model)

Class for accessing information from model behaviour at each observation"
class skika.evaluation.inspect_recurrence.DataExposure(data)#

Class for accessing information from data at each observation”

Parameters

data – An object containing data information at all observations.

class skika.evaluation.inspect_recurrence.InspectPrequential(max_samples=100000, pretrain_size=200, output_file=None, show_plot=None, data_expose=None, name='test.pdf')#

Class to run classifier and pass information to the visualizer. Iteritively runs classifier on each data stream observation and extracts measures to plot.

Parameters

  • max_samples (int) – The maximum number of observations to run.

  • pretrain_size (int) – The number of observations to run before starting to plot. To reduce unstable results at low observation count.

  • output_file (str) – The output file name.

  • show_plot (bool) – Whether or not to show a real time plot. Much slower.

  • data_expose – An object containing information about the datastream, for example drift points.

  • name (str) – Name of the experiment.

class skika.evaluation.inspect_recurrence.InspectorVisualizer(name='test.pdf')#

Class to visualize and plot online classification results in real time. Specific data is exposed per train step, and ploted on a real time graph.

Parameters

name (str) – Name of output files

hold()#

Function called at end of run, can save to output file or hold for viewing.

on_new_train_step(sample_id, data_exposure, model_exposure, X, y, p)#

Called every training step. Exposed data is read and plotted.

Parameters

  • sample_id (int) – The number of the incoming data stream observation.

  • data_exposure – Object containing information to plot from data.

  • model_exposure – Object containing information to plot from model behaviour.

  • X – Observation input

  • y – Observation label

  • p – Model predition for observation

class skika.evaluation.inspect_recurrence.ModelExposure(model)#

Class for accessing information from model behaviour at each observation”

Parameters

model – An object containing model information at all observations.