适用于iPhone / iPad的线性回归库或代码段？

Question

I would like to find the line that best fits n cartesian points.

Is there a package or library that will help me on iPhone - something in C or objective-C?

Answer 1

You can write a quick code to do that. The least-squares solution to the problem is b = (G^(T)G)^(-1)G^(T)m, where b is the vector of coefficients of the linear fit, m is the vector of all y values and G is the kernel matrix of dimension (length(m),2) and each row of G is (1, x) where x is the x value corresponding to ay value. G^T is the transpose of G.

The transpose and inversion can be done by using LAPACK library (or any other linear algebra library of your choice) or by taking the appropriate code from a Numerical Recipes book.