I must admit, I have never seen nor used an SVD (Singular Value Decomposition) to fit a line to a data set before. In order to overcome this unfamiliarity, I began by computing an SVD by hand to a small matrix and learning along the way (from Neil's notes) that a rectangular matrix can be broken down into a product of three matrices:

Translating the concept into Python for generating 100 points between 0 and 1, letting y = 2 + 3x + g where "g" is a gaussian random variable, yields:

Attempts at knowing a data set drawn from a Gaussian distribution to find an optimal estimator of the mean (with the help of fellow NMM classmates):

(Some) References Online

http://terpconnect.umd.edu/~toh/spectrum/CurveFitting.html

http://docs.scipy.org/doc/numpy/reference/generated/numpy.linalg.svd.html

http://docs.scipy.org/doc/numpy/reference/generated/numpy.linalg.svd.html

http://pingswept.org/2009/01/24/least-squares-polynomial-fitting-in-python/