Let's say we have 10 possibly good predictor variables and one Y variable. We want to make a model for each possible combination of predictors (for example, there would be ten models of just each predictor variable alone, with an intercept). This way, we can see what the best combination of variables is. How many models would we have to create to create every possible model? Assume we also make a model of just the intercept (which is the assumption made in the book and slides when talking about this).