On 23.05.2012 14:57, terriblement wrote:
> I've a file with>100 000 points, which describe an exponential function,
> but as the curve is not smooth at all, the fit doesn't match.
That conclusion is most probably invalid. Neither the sheer number of
data points, nor the data being noisy, will cause the fit not to
"match", just like that.
More likely 1) the original assumption that an exponential will match
that data is wrong, or 2) you're hitting a technical limit of gnuplot's
current implementation of 'fit'.
|