python implementation of Pavel Holoborodko's method of "Smooth noise-robust differentiators"
This routine will take a function as defined by data points as input and return an approximation to the first derivative of the function. See http://www.holoborodko.com/pavel/numerical-methods/numerical-derivative/smooth-low-noise-differentiators