Weibull fit to interval censored data

I am trying to get the best weibull fit to some interval censored data. I know that wblfit function can handle right censored data, but it cannot handle interval censored data. Any help would be appreciated.

Answer 1

[Jeff Miller]

You can do this with Cupid. Here is an example:

% Generate some truncated Weibull data just to have an example
% of some data to fit:
SampleSize = 1000;
TrueOffset = 0;
TrueScale = 1.3;
TrueShape = 1.2;
x = wblrnd(TrueScale,TrueShape,SampleSize,1) + TrueOffset;

% Here are the truncation limits & truncated data.
LowerLim = 0.5;
UpperLim = 4;
xt = x(x>LowerLim & x<UpperLim);

% Set up a doubly-truncated Weibull in Cupid:
GuessOffset = 0;
GuessScale = 1.4;
GuessShape = 1.1;
w = Weibull(GuessScale,GuessShape,GuessOffset);
tw = TruncatedX(w,LowerLim,UpperLim);

% The next line will use xt to compute maximum likelihood estimates of the
% three Weibull parameters scale, shape, and offset
tw.EstML(xt,'rrrff')  % The optional string 'rrrff' indicates that the first 3 parameters are
                      % free to vary as reals, and the last 2 parameters are fixed.
                      % The first 3 parameters are those of the Weibull, and the last two
                      % parameters are the lower/upper limits

% If you want to fix the offset, just change the third r to an f like this: 
tw.EstML(xt,'rrfff')