How to fit multiple curves to histogram subpopulations?

Hello, I will be as succint as possible. I have a histogram and would like to fit two seperate curves to the distribution, representing the two subpopulations in the distribution. Essentially I would like to do this (red lines I drew):

How can I do this?

Note that these arent two seperate histograms, but rather a single histogram.

My code for this histogram is here:

%%
figure;
d1=histogram(dG_TfR_100, [0: .3: 25], 'Normalization' ,'probability'); %range of current drops 
ylabel('normalized event count')

I have also attached the variable dG_TfR_100 as an excel sheet in case it is needed.

Please let me know if there is any additional information you need. Thank you.

Answer 1

[AndresVar]

You could fit a distribution to the histrogram using histfit or fitdist, then identify the number of modes or peaks, N. Then fit the distribution with the sum of N distributions, with carefully selected parameters:

Histogram with a distribution fit - MATLAB histfit (mathworks.com)

Kernel probability distribution object - MATLAB (mathworks.com)

Find local maxima - MATLAB findpeaks (mathworks.com)

Curve Fitting - MATLAB & Simulink (mathworks.com) to fit your choice of components

You can try this method that fits gaussian mixture model, it works like a k-means clustering:

Fit Gaussian mixture model to data - MATLAB fitgmdist (mathworks.com)

Browse file exchange for helpful functions:

1D: gaussian_mixture_model.m - File Exchange - MATLAB Central (mathworks.com)

Flexible mixture models for automatic clustering - File Exchange - MATLAB Central (mathworks.com)

Here is a gaussian decomposition paper

(PDF) Gaussian Decomposition of Multimodal Curve and its Application to Sedimentology (researchgate.net)