How to fit a mixture of uniform and gaussian distributions

Hi there, Does anyone know / have any tips about / how to fit a mixture of (1) a uniform distribution (1 height parameter) and a (2) Gaussian? I know how to fit a mixture Gaussian... but am flummoxed re how to go about fitting a mixture involving a uniform and an gaussian. Many thanks in advance for your help, Marie

Answer 1

[Salaheddin Hosseinzadeh]

Hi Maria,

What is a uniform distribution? something like a DC offset? Is it like a constant added to the Gaussian?

See if you have curve fitting toolbox

cftool

you can define a custom equation gauss + constant and get the parameters you want.

Answer 2

[Marie.a.londres]

Hi there, Many thanks - a uniform distribution can described as a rectangle (below the Gaussian) so it will have a height parameter (it could have a dispersion too, I suppose, but not of interest in the case I am attempting to tackle)... I can't model it as a constant in this case as it will vary (theoretically) from one condition to another in the data I have. So if you think of the uniform as a sort of added even noise, there will be more noise in some conditions and less in others. I do have the curve fitting toolbox - still working on getting my head around the code for custom distribution fitting. Cheers - Marie

[Salaheddin Hosseinzadeh]

Hi Maria,

Try the exponential fitting in cftool