Model Complexity
Model complexity may seem like a strange title for a chapter. Why should we care about complexity? One concept you may have already encountered as a data scientist is that of overfitting and how an overfitted model will not make accurate predictions. However, that overfitting stems from using a model whose complexity is greater than that justified by the data. The impact of model complexity on model prediction accuracy is a nuanced one. More specifically, how you decide what is the right level of model complexity can be challenging. To address this challenge requires exploring several new concepts. We will do that exploration in this chapter and do so by covering the following topics:
- Generalization, overfitting, and the role of model complexity: Here, we understand how model complexity affects the accuracy of model predictions on unseen data
- The bias-variance trade-off: Here, we dig into the mathematical details behind the prediction accuracy ideas we...
