I am trying to approximate a multivariate function $y = f(x_1, ...x_n)$, which I have reason to believe will be well approximated by a classification and regression tree. Some of the variables are categorical, most of the variables are numeric. The target variable is numeric.
The size of the problem is going to be an issue for me: I have 2 billion observations and 300 input variables ($n = 300$ in the above).
Other than downsampling until my dataset fits in memory and/or the CART algorithm terminates in reasonable time, what are my options?
