Is it possible to create a distribution fit (preferably gaussian mixture distribution model) for a distribution with precalculated bincounts?

The same way you can use the edges & bincounts of the bins as input to the histogram function instead of the data array.

Every function I tried (histfit, fitdist, fitgmdist) seems to necessarily need the actual data as input. The reason I can't do this is that I want the distribution bins to be weighted with different factors.

Background:

I have a data set of concrentration & mean radio carbon age for different types of carbon compounds. I want to create a histogram of the age distribution, but concentration weighted. So, when adding the mean age of a type of compound to the histogram, the bin is supposed to have the height of the concentration of that type (rather than just heigtht 1 for 1 count in the data set).

Answer 1

[Torsten]

The bincounts define a probability distribution. Draw random numbers from this distribution - they can be used as data in "fitdist".