Classification Learner App Iterations

Hi,

I am performing supervised learning on a binary data set of 20 samples. I am doing holdout variation, training on 80% and testing on 20% of the data. As 20% of the test data is only 4 samples, the accuracy I get out each time I run a classifier varies wildly. I have found a way to manually add repeat iterations by generating the code from the app and adding it in (I am currently using 1000 iterations). However, I would like to be able to use the optimisable classifiers found in the app, but they would only be valuable if I can find a way to add repeat iterations using different holdout samples for each run. Any help would be much appreciated!

Answer 1

[Prince Kumar]

Hi,

Your training sample is too small for the model to learn anything significant. If you train your model for large interations, then your model will overfit.

Hope this helps!