[Quantlib-users] Interpolation/Smoothing Algo for 2D surface

[Michael (D. portal) <mi...@da...>]

Hi All,

I have a 2D matrix with each cell that is backed by observations with
certain characteristics and shows weighted average performance values (see
below for an example).

Some of the cells are backed by too few observations and therefore this 2D
surface is not smooth enough in those cells (see yellow cells below e.g.)

What is the best algo for smoothing this surface taking into account the
number of observations (e.g. cells with more observations should we
weighted heavier). E.g. is there a Python weighted spline lib available or
something similar?

[image: image.png]

Thanks,

Michael