In this section, we will see supervised learning in action. We won't look at complicated algorithms, but we will look at how to train even a simple algorithm and machine learning best practices, such as splitting data into training and test data, and performing cross-validation.
Training models
Issues in training supervised learning models
When a model does not predict the target variable well, it underfits. This is true for both seen and unseen future data. Underfitting is when an algorithm trained to predict a value does so poorly both on the training data and on future, unseen data. Overfitting is when a model predicts training data well, but does not generalize well, and so predicts future data poorly. Data analysts...
