# Minimum number of samples to train XGBoost without overfitting

When using Neural Networks for image processing I learned a rule of thumb: to avoid overfitting, supply at least 10 training examples for every neuron.

Is there a similar rule of thumb for classifiers such as XGBoost, presumably taking into account the number of features and estimators?

And, considering the 'curse of dimensionality' shouldn't the rule of thumb be that n_training is geometric in n_dimensions, and not linear?

• Or by other parameters that reduce overfitting of XGBoost, for example minimum weight to create split (min_child_weight in XGBoost package in Python). – Itamar Mushkin Jan 19 '20 at 6:19