ROC curve

Summary

• Plugin name : ROC curve
• Version : 0.1
• Author : Serrière Guillaume
• Company : Inria
• Short description : Compute the ROC curve for each class.
• Documentation template generation date : Jan 9 2018

Description

The box computes the ROC curve for each class.

This box computes the ROC (Receiver Operating Characteristic) curve for a classifier (https://fr.wikipedia.org/wiki/Receiver_Operating_Characteristic). One curve will be computes by class. This box is designed to work with the probability output of the Classifier processor.

The box will compute the curve when it receives the computation trigger on the first input.

The result is displayed when computed in a standalone visualization.

Inputs

1. Expected labels

The first input receives the expected stimulations stream.

• Type identifier : Stimulations (0x6f752dd0, 0x082a321e)

2. Probability values

This input receives the probability output stream of the processor box.

• Type identifier : Streamed matrix (0x544a003e, 0x6dcba5f6)

Settings

Each setting after the second one corresponds to the stimulation code of a class. A stimulation must be unique.

1. Computation trigger

Stimulation trigger for the computation.

• Type identifier : Stimulation (0x2c132d6e, 0x44ab0d97)
• Default value : [ OVTK_StimulationId_ExperimentStop ]

2. Number of classes

This setting indicates the amount of classes handled by the box. This setting will change the amount of setting.

• Type identifier : Integer (0x007deef9, 0x2f3e95c6)
• Default value : [ 2 ]

3. Class 1 identifier

This setting indicates the stimulation corresponding to the first class.

• Type identifier : Stimulation (0x2c132d6e, 0x44ab0d97)
• Default value : [ OVTK_StimulationId_Label_01 ]

4. Class 2 identifier

This setting indicates the stimulation corresponding to the second class.

• Type identifier : Stimulation (0x2c132d6e, 0x44ab0d97)
• Default value : [ OVTK_StimulationId_Label_02 ]

Examples

Miscellaneous

All stimulations can be send to the box. They will be filtered.