You've trained a model that looks very strong on the training set, but the team is worried it may not generalize well to new data. You want to verify whether the model is learning real signal or memorizing patterns from the training sample.
How would you investigate whether a model is overfitting?
Large gap between training and validation metricsCross-validation performance consistently below training performanceValidation log loss much worse than training log lossModel complexity that improves train fit without improving holdout results