prediction

Module: environments.live.prediction

Script to run the actual online classification of data

Inheritance diagram for pySPACE.environments.live.prediction:

Class Summary

SimpleResultCollection(name, params[, store])	Base Class for Result collection.
ConfusionMatrix(name, params)	A confusion matrix.
Predictor([live_processing])	Class that is responsible to perform the actual predictions.

Classes

SimpleResultCollection

class pySPACE.environments.live.prediction.SimpleResultCollection(name, params, store=True)

Bases: object

Base Class for Result collection.

Default behaviour is counting occurring events in a dictionary.

Parameters

name:potential-name (e.g. LRP) params:parameter/meta-data of the potential store:write result to a pickle file

Class Components Summary

__repr__()
dump()	simple print function
event_notification(event_str)	simple event counting

__init__(name, params, store=True)

__repr__()

event_notification(event_str)

simple event counting

dump()

simple print function

__weakref__

list of weak references to the object (if defined)

ConfusionMatrix

class pySPACE.environments.live.prediction.ConfusionMatrix(name, params)

Bases: pySPACE.environments.live.prediction.SimpleResultCollection

A confusion matrix.

Stores and handles a confusion matrix. The confusion matrix is assumed to have the following form:

		prediction
		P	N
actual	P	TP	FN
	N	FP	TN

Parameters

name:potential-name (e.g. LRP) params:parameter/meta-data of the potential

Class Components Summary

__repr__()
dump()	prints the collected result of the confusion matrix
event_notification(event_str)	Update the confusion matrix

__init__(name, params)

__repr__()

event_notification(event_str)

Update the confusion matrix

It is assumed that we receive a trigger event after the classification result (either pos. or neg.-event). If the trigger_event occurrs it is a validation for a positive prediction. If no trigger_event but instead another classification result comes in it is assumed that no reaction on the previously presented target appeared.

dump()

prints the collected result of the confusion matrix

Predictor

class pySPACE.environments.live.prediction.Predictor(live_processing=None)

Bases: object

Class that is responsible to perform the actual predictions.

Class Components Summary

__del__()
create_processor_logger()	Create specific logger for the prediction
initialize_xmlrpc(controller_host, ...[, ...])	Setup communication to remote listeners
is_predicting_active()	Returns whether prediction phase is finished or still running
load_model(directory, datasets)	Store the learned models
pause_prediction()
prepare_predicting(datasets[, testing_data, ...])	Prepares the trained aBRI-DP flows to classify new instances.
process_external_command(command)
set_controller(controller)	Set reference to the controller
set_eeg_stream_manager(stream_manager)	Set manager class that provides the actual data for the prediction
start_predicting([trace])	Classify new instances based on the learned aBRI-DP flows.
stop_predicting()	Force the end of the predicting

__init__(live_processing=None)

__del__()

initialize_xmlrpc(controller_host, controller_port, mars_host='127.0.0.1', mars_port=8080)

Setup communication to remote listeners

This method tells ABRIProcessing which remote processes are interested in being informed about its classification results.

set_eeg_stream_manager(stream_manager)

Set manager class that provides the actual data for the prediction

load_model(directory, datasets)

Store the learned models

prepare_predicting(datasets, testing_data=None, nullmarker_stride_ms=None)

Prepares the trained aBRI-DP flows to classify new instances.

start_predicting(trace=False)

Classify new instances based on the learned aBRI-DP flows.

is_predicting_active()

Returns whether prediction phase is finished or still running

process_external_command(command)

pause_prediction()

stop_predicting()

Force the end of the predicting

set_controller(controller)

Set reference to the controller

__weakref__

list of weak references to the object (if defined)

create_processor_logger()

Create specific logger for the prediction