Hello everyone. I need your help. I'm trying to make a loop. In this loop which is 10 iterations, as it is within a cross-validation, I must first make a division between the two classes that are required to predict. For this I need a random seed, which should be changing in each loop so that the cross-validation rerun the same results. Next I need to separate using a Row Splitter both classes, since I randomly select the same number of minority class in the ruling class. This also needs another random seed to behave the same way as the start of the cross-validation. Then along all with a Concatenate Node. The point is that in each sample, the same amount of records does not appear in both classes, therefore, for the second sample in the Raw Sampling, I need to know the number of records that appear on the minority class to thereby obtain the same amount of the other kind. The reason for this flow is I'm doing a predictive model for a database in which the variable to be predicted is unbalanced, there is a 14% of a class and 86% other. The objective is to present the model loop 10 bases of training data that are balanced. I add the flow I've done in KNIME_project3.
Your comments and support will be very well received.
Thank you very much
Gabriel Cornejo
CHILE