Improving generalization through model validation

Why does this matter?

An efficient and robust model performs well not only on the training set, but also on unseen data. The process of testing a trained model on unseen data is called model validation. The ability of a model to make accurate predictions on unseen data, is called generalization.

It is therefore important to employ appropriate techniques to evaluate how well an algorithm generalizes. This will also help us select the best-performing hyperparameters and even the right machine learning algorithm for our problem.

This Roadmap

This guide will walk you through a number of techniques for model validation. Firstly, we will cover techniques to evaluate the performance of the model on unseen data. Secondly, we will cover techniques for tuning the hyperparameters of the model. Finally we will cover how to perform both algorithmic and model selection together.

Last updated