divideind

(To be removed) Divide targets into three sets using specified indices

divideind will be removed in a future release. For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.

For advice on updating your code, see Version History.

Syntax

[trainInd,valInd,testInd] = divideind(Q,trainInd,valInd,testInd)

Description

[trainInd,valInd,testInd] = divideind(Q,trainInd,valInd,testInd) separates targets into three sets: training, validation, and testing, according to indices provided. It actually returns the same indices it receives as arguments; its purpose is to allow the indices to be used for training, validation, and testing for a network to be set manually.

The indices are returned after removing any indices greater than Q. Note that some indices in the range 1:Q may not be assigned to any of the three sets, and the same indices should not be used in more than one set.

Examples

Divide Samples into Three Sets Using Specified Indices

This example shows how to divide samples into three sets using specified indices for a network.

Divide 20 samples into training, validation and test indices, so that only 16 are actually used.

[trainInd,valInd,testInd] = divideind(20,1:8,9:12,14:16)

This code shows you how to ensure a network performs the same kind of data division when it is trained:

net.divideFcn = 'divideind';
net.divideParam.trainInd = 1:8;
net.divideParam.valInd = 9:12;
net.divideParam.testInd= 14:16;

Input Arguments

Q — Number of targets

scalar

Number of targets to divide up, specified as a scalar.

trainInd — Training indices

vector

Training indices, specified as a 1-by-Q vector.

valInd — Validation indices

vector

Validation indices, specified as a 1-by-Q vector.

testInd — Testing indices

vector

Testing indices, specified as a 1-by-Q vector.

Output Arguments

trainInd — Training indices

vector

Training indices, returned as a vector.

valInd — Validation indices

vector

Validation indices, returned as a vector.

testInd — Testing indices

vector

Testing indices, returned as a vector.

More About

Network Use

These are the network properties that define which data division function to use, what its parameters are, and what aspects of targets are divided up, when train is called.

net.divideFcn
net.divideParam
net.divideMode

Version History

Introduced in R2008a

R2026a: divideind will be removed

divideind will be removed in a future release.

To learn more about data processing for neural networks, see these examples and topics:

• Preprocess Data for Domain-Specific Deep Learning Applications

• Datastores for Deep Learning

• Choose an App to Label Ground Truth Data

For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.