# How to fit a dataset with two independant variables (of different dimensions) and one dependant variable?

15 views (last 30 days)
Ashok Jacob on 15 Mar 2021
Answered: Jeff Miller on 16 Mar 2021
I need to fit data points of an Intensity-Duration-Frequency curve into a non-linear equation of the form
i=K*(T^a)/((t+b)^n)
where i is the intensity(dependant),T the return period(independant) and t the duration(independant).
K,a,b,n are the coefficients to be found.
t={ 1 2 4 5 10 12 15 24 36 48 }
T={2 5 10 25 50 100}
I was attempting nlinfit with the two independant variable data sets to be in one single matrix,but dimensions do not agree for the two.How would i fit these?
##### 2 CommentsShowHide 1 older comment
Ashok Jacob on 16 Mar 2021
there are 10*6= 60 data points in i,one for each combination of t and T.
Apologies for the oversight.

Jeff Miller on 16 Mar 2021
You need give nlinfit the corresponding 60 values of t and T. Something like this (but check the order of the resulting 60 points):
t=[ 1 2 4 5 10 12 15 24 36 48];
T=[2 5 10 25 50 100];
[t2, T2] = ndgrid(t,T);
t2 = reshape(t2,60,1);
T2 = reshape(T2,60,1);
% now pass t2 and T2 to nlinfit

R2015a

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!