I'm surveying some papers for different kinds of approaches for (statistical) function modelling.

Suppose (x_1,...,x_n) is one of multiple input data points and its output value is y.

An intuitive idea for modelling the function is f(x_1,...,x_n)=w_1*x_1+...+w_n*x_n, which is a weighted sum. However, the output value of the weighted sum function may not be close to y enough.

So we want to find some function F, which can be nonlinear, such that its output value F(x_1,...,x_n)is very close to y.

The goal of paper survey is to survey different kinds of approaches for (statistical) function modelling and what problems are suitable for each kind of function model.

Every suggestion for survey would be greatly appreciated.