Building data and performing least squares regression with cosines in java

Question

I have data that I would like to build, and, more importantly, least squares regression when using cosines (instead of using polynomials):

Any recommendations? Thanks.

Answer 1

Probably the following page solves the regression part of your goal:

http://www.teneighty.org/software/index.html?f=fft&c=e98b8

Answer 2

You can find this demo “Least Squares and Fitting Data” as it solves some of your problems.

Answer 3

Just a bit of cautionary advice. Using the Fourier series makes sense if you think that your basic function has a cosine series as the basis; however, if you use it as the basis for an arbitrary function (with an unknown shape), you can try to guess a more specific basic type of function (polynomial, exponential, etc.).

I did some limited optimization on such a series, and the function was so distorted that it was hard to say if my fit was significant; my matching function had a large number of local maxima.

Answer 4

MathGL can be built, placed (using GSL) and show the result of the finish - see this sample.