# how to prolong the fit distribution curve to reach x-axis ?

1 view (last 30 days)
Amr Hashem on 31 Oct 2015
Commented: Star Strider on 1 Nov 2015
I got the best fit distribution to a failure rate (data)
Now I want to prolong the fitted distribution curve to reach x-axis (in both tails)
I use:
binWidth = 1;
binCtrs = 0:binWidth:18;
figure
hist(data,binCtrs);
counts = hist(data,binCtrs);
nu = length(data);
prob = counts / (nu * binWidth); % rescale the figure
figure
plot(binCtrs,prob,'ro');
paramEsts = gevfit(data); %estimate parameter
xgrid = linspace(0,18,100);
pdfEst = gevpdf(xgrid,paramEsts(1),paramEsts(2),paramEsts(3)); hold on;
line(xgrid,pdfEst,'color','b')
to (fit distribution &) produce this figure:
Now , I want to prolong/extend the curve to reach x-axis
I try:
xaxis=linspace(-5,22); % create Matching vector
line(xaxis,pdfEst,'color','r')
but it produced another distribution.
Amr Hashem on 31 Oct 2015
Any suggestion...!

Star Strider on 31 Oct 2015
According to the documentation on the Generalized Extreme Value Distribution particualrly the Examples section, it may or it may not have a zero probability at some value of x. (It will never cross zero.) It all depends on what the distribution is for your data. Extend your xgrid vector in the negative and positive directions and see what it does.
Experiment!

Amr Hashem on 31 Oct 2015
I got it...
xgrid = linspace(-20,20,100);
Star Strider on 1 Nov 2015
There you go!