The SUMO toolbox  2018a
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Pages
Public Member Functions | List of all members
ValidationSet Class Reference

This measure uses a set of validation samples, not used for construction of the model, to estimate the accuracy of the model. More...

Inheritance diagram for ValidationSet:
Inheritance graph
[legend]

Public Member Functions

function ValidationSet (var varargin)
 Class constructor. More...
 
function calculateMeasure (var model, var context, var outputIndex)
 Calculates the measure. More...
 
function calculateValidationSet (var samples)
 Creates a validation set from the full set of training samples. More...
 
function setErrorFcn (var efun)
 Sets the error function to use. More...
 
function getErrorFcn ()
 Returns the used error function. More...
 
function getFinalTarget ()
 Returns the target (goal) accuracy. More...
 
function getParallelMode ()
 Returns whether we are using parallel computing. More...
 
function getWeight ()
 Returns vector of weights. More...
 
function isEnabled ()
 Returns whether this measures is being used. More...
 
function getTarget ()
 Returns the target (goal) accuracy. More...
 
function processMeasure (var model, var context, var outputIndex)
 Wrapper that calls calculateMeasure. More...
 

Detailed Description

This measure uses a set of validation samples, not used for construction of the model, to estimate the accuracy of the model.

There are two sources for this set of validation samples:

This measure is very useful for validation purposes.

Constructor & Destructor Documentation

function ValidationSet ( var  varargin)
inline

Class constructor.

Returns
instance of class
Todo:
ValidationSet works only inside SUMO because of SampleManager

Member Function Documentation

function calculateMeasure ( var  model,
var  context,
var  outputIndex 
)

Calculates the measure.

Splits the list of samples in a set of validation samples and a set of training samples.

Parameters
modelsurrogate model
contextstruct of contextual information (optional)
outputIndexfor which output to calculate the measure
Return values
newModelupdated surrogate model (may happen in some measures)
scorescore of the surrogate model

Then a new model is constructed using the training samples, and the accuracy of this model is validationed using the validation samples. When an external dataset is provided, the entire list of samples is used for training and the model is evaluated against the dataset. The model that was constructed is returned, so that sub-measures can act on this model.

function calculateValidationSet ( var  samples)

Creates a validation set from the full set of training samples.

Parameters
samplestraining set
function getErrorFcn ( )
inlineinherited

Returns the used error function.

function getFinalTarget ( )
inlineinherited

Returns the target (goal) accuracy.

function getParallelMode ( )
inlineinherited

Returns whether we are using parallel computing.

function getTarget ( )
inlineinherited

Returns the target (goal) accuracy.

function getWeight ( )
inlineinherited

Returns vector of weights.

function isEnabled ( )
inlineinherited

Returns whether this measures is being used.

function processMeasure ( var  model,
var  context,
var  outputIndex 
)
inherited

Wrapper that calls calculateMeasure.

This function calls calculateMeasure and replaces NaN/Inf Values.

Parameters
modelsurrogate model
contextstruct of contextual information (optional)
outputIndexfor which output to calculate the measure
Return values
scorescore of the surrogate model
function setErrorFcn ( var  efun)
inlineinherited

Sets the error function to use.


The documentation for this class was generated from the following files: