Regularizing seasonality
Often, when solving a problem with machine learning, the data involved is so complex that a simple model just isn't powerful enough to capture the full subtlety of the patterns to be found. The simple model tends to underfit the data. In contrast, a more complicated model with many parameters and great flexibility can tend to overfit the data. It is not always easy, or possible, to use a simpler model. In these cases, regularization is a good technique to use in order to control overfitting.
Prophet is such a powerful forecasting tool that without care, it can sometimes be very easy to overfit the data. That's why understanding Prophet's regularization parameters can be quite useful.
Tip
A model is said to be underfit if it does not fully capture the true relationship between the input features and the output features. Performance is low on both the training data and any unseen testing data.
A model is said to be overfit if it goes...
