[Numpy-discussion] nan result from np.linalg.lstsq()

[Larry Paltrow]

I'm having some trouble using the linalg.lstsq() function with certain data
and trying to understand why. It's always returning nans when I feed it
this numpy array of float64 data:

data = df.close.values #coming from a pandas dataframe
type(data)
>>> numpy.ndarray
data.dtype
>>> dtype('float64')
data
>>> array([ 1.31570348, 1.31565421, 1.3157375 , ..., 1.32175 ,

        1.32180441,  1.321775  ])


xi = np.arange(0,len(data))
A = np.vstack([xi, np.ones(len(xi))]).T A
>>> array([[ 0.00000000e+00, 1.00000000e+00],

       [  1.00000000e+00,   1.00000000e+00],
       [  2.00000000e+00,   1.00000000e+00],
       ...,
       [  2.87800000e+03,   1.00000000e+00],
       [  2.87900000e+03,   1.00000000e+00],
       [  2.88000000e+03,   1.00000000e+00]])


m, c = np.linalg.lstsq(A, data)[0]
m, c
>>> (nan, nan)

oy, do not understand. Does anyone else know why?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/numpy-discussion/attachments/20121029/956dfac0/attachment.html