How can I estimate parameters of the student's t distribution in a loop?

I have a loop which generates several random numbers, each time I have to estimate their parameters using the univariate Student's t and then use the estimated parameters later on. Is there any way I can do this? fitdist lets me estimate them however it doesn't seem to let me 'call' them so I cannot use them later on in the mfile.

[Star Strider]

What parameters do you want to estimate for the random vector generated by each iteration of the loop? Mean and standard deviation? T-statistic and degrees of freedom for each element in the vector? Confidence interval for the the vector? Something else?

[May]

mean, standard deviation and degrees of freedom

[dpb]

Well, now are you talking about the parameters of the fitted t or those of the random sample?

As noted in my answer, the fit results are in the named structure fields mu and sigma returned by fitdist; the DOF is simple to compute given there are two estimated parameters.

[May]

my apologies if I'm not being clear enough. I'm not very familiar with matlab - basically I'm trying to compute the following

fitdist(uu1','tlocationscale')

x1=icdf('t',uu1,nu,mu,sigma)

Matlab returns this

ans =

tlocationscale distribution

    mu = 0.527668
    sigma = 0.290663
    nu = 9.58997e+006

??? Undefined function or variable 'nu'.

Which is what I meant by matlab doesn't let me 'call' the parameters. I cant just copy and paste the values in the formula of x1 as I need to do this multiple times (in a loop - uu1 in a random generated vector)

[dpb]

x1 is a structure whose field names are mu, sigma, nu.

To reference them, it's

x1.mu

etc. Read in the doc of "data Types" for more details of Matlab.

[May]

thank you!!

Answer 1

[dpb]

The returned distribution object (structure) has fields mu and sigma