Cross validation in matlab

6 views (last 30 days)
Lester Lim
Lester Lim on 30 Jan 2013
Edited: Greg Heath on 1 Jan 2018
What are the steps to performing cross validation on labels of data to get the accuracy of the results?

Accepted Answer

Greg Heath
Greg Heath on 30 Jan 2013
Edited: Greg Heath on 1 Jan 2018
Repeat until the parameter estimates converges
1.Randomly divide the data into 10 subsets
2.For each subset
a. Use the remaining 9 subsets to design a model
b. Test the model with the holdout subset
c. Update the average and standard deviation of
the holdout test set error.
d. If std < thresh1 or std < thresh2*avg, stop.
Hope this helps.
Thank you for formally accepting my answer.
Greg

More Answers (1)

Ilya
Ilya on 30 Jan 2013
The Statistics Toolbox provides utilities for cross-validation. If you are using R2011a or later, take a look at ClassificationTree.fit, ClassificationDiscriminant.fit, ClassificationKNN.fit and fitensemble. Notice the 'crossval' parameter and other related parameters. If you are working in an older release or not using any of these classifiers, the crossval function is a generic utility for that purpose.
  1 Comment
Lester Lim
Lester Lim on 31 Jan 2013
Is there any other way to do it without stats toolbox?

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!