GAMs are great tools. However in my understanding they can be quite unstable when working with (subsets of) medium sized data. In other words: it's easy to formulate quite complex models that seem to work, but fail at prediction. So as with any complex modelling mastery is the key.
GAMs are great tools. However in my understanding they can be quite unstable when working with (subsets of) medium sized data. In other words: it's easy to formulate quite complex models that seem to work, but fail at prediction. So as with any complex modelling mastery is the key.