x-partitioner partitions size

Hello,

The default partitions size in x-partitioner is about 67% training and 33% testing. How can I choose different partitions sizes? Thanks a lot for any help.

Hussam

Hello hghumaim,

The x-partitioner node allows to set the number of cross validation iterations that should be performed. The default number of validations is set to 10.¬†In¬†k-fold cross-validation, the original sample is partitioned into¬†k¬†equal sized subsamples. Of the¬†k¬†subsamples, a single subsample is retained as the validation data for testing the model, and the remaining¬†k¬†‚ąí¬†1 subsamples are used as training data. This means that if you set the number of validations as 10, you will have 10% used for validation and 90% used for training in each iterationl. Of course, the cross-validation process is then repeated¬†k¬†times (the¬†folds), with each of the¬†k¬†subsamples used exactly once as the validation data.¬†

The parameter "number of validations" set in the configuration node will changethe percentage of your data used in each iteration, both for the validation set and the training set.

What your 67% and 33% are referring to exactly?

Best,

Vincenzo

1 Like