number_of_training_instances/number_of_training_features ratio



What ratio of number_of_training_instances vs. number_of_training_features is a good thumb rule for getting a good fit machine learning model?

Say we have 10 features and 10 training instances, i.e. ratio is one. We will not get a good fit model.
What ratio can I use as a thumb rule to validate datasets before training?


@Bolaka-with less training data, your parameter estimates have greater variance.
so it good that ratio should 4:1 or 5:1

Hope this helps!