Science these days is about making models, not simple generalizations. Models are tested by overall fit rather than by specific predictions. And they are designed to offer a specific perspective (often incorrectly called a 'lens') on a domain. But they can constrain our thinking, making it difficult to imagine any possible state of affairs not described by the model. But in models as in all things, diversity is better. "While applying one model is good, using many models — an ensemble — is even better, particularly in complex problem domains. Here's why: models simplify. So, no matter how much data a model embeds, it will always miss some relevant variable or leave out some interaction. Therefore, any model will be wrong." I would add that it's really hard for a single person to make many models, which is why a collection of people independently creating models offers the diversity we need without the inhetrent difficulty it otherwise entails.
Today: 3 Total: 3 [Direct link] [Share]
