An interesting, arduous, and repetitive part of machine learning is the model evaluation journey. There is again, art and a different mindset required to build models that are robust. Throughout this book, we have simplified the model evaluation process with training and testing datasets that were derived by splitting the available data into a 70:30 or 80:20 ratio. Although this approach was effective in helping us understand how the model performs on unseen data, it still leaves several loopholes that might render the model futile for most other cases. We will need a more formal, thorough, and exhaustive method of validation for a machine learning model to be robust for future prediction events. In this chapter, we will study cross-validation and its various approaches to assess the performance of a machine learning model.
Before we delve into the specifics of the topic, we need to explore a crucial topic in machine learning called bias-variance trade-off. This topic...