you could split with uniform age ranges or equally-distributed https://www.vision.ee.ethz.ch/en/publications/papers/articles/eth_biwi_01299.pdf here is written the last one gives better results check if it can be done in our given time you have to talk with the nn groups if they use regression or classification ## usefull links: - http://www.scikit-yb.org/en/develop/api/target/class_balance.html - https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.StratifiedShuffleSplit.html - https://imbalanced-learn.org/en/stable/index.html
you could split with uniform age ranges or equally-distributed
https://www.vision.ee.ethz.ch/en/publications/papers/articles/eth_biwi_01299.pdf
here is written the last one gives better results
check if it can be done in our given time
you have to talk with the nn groups if they use regression or classification
usefull links: