Error using gamfit . X must be non-negative.

clear;clc
%hist = 'access05_indo_daily.txt';     % as x
%obs = 'chirps05_indo_daily.txt';   % as y
hist = 'hfeb_.txt';     % as x
obs = 'ofeb.txt';   %  as y
delimiterIn = ' ';
%headerlinesIn = 1;
x = importdata(hist);
y = importdata(obs);
[m,n] = size(y);
ycor = zeros(m,n);
%cpolys = zeros(n,4);
    xi = x(:,4);
    yi = y(:,4);
    yi(yi<=0) = 0;
    Pxi = gamfit(xi);
    Pyi = gamfit(yi(~isnan(yi)));
    xgrid = linspace(0,1,101);
    xinv = gaminv(xgrid,Pxi(1),Pxi(2)); xinv = xinv(1:100); % nilai ke 101 biasanya infinite value, jadi harus dibuang
    yinv = gaminv(xgrid,Pyi(1),Pyi(2)); yinv = yinv(1:100); % nilai ke 101 biasanya infinite value, jadi harus dibuang
    cpoly = polyfit(xinv,yinv,4); 
    %cpolys(i,:) = cpoly;

[Arly Getha Purba]

What should I do?

[the cyclist]

Can you upload your input files, so that we can run your code and see what is going on?

[the cyclist]

@Arly Getha Purba, I saw that you emailed a link to the file, but I'm not going to open a file I can't scan. Sorry.

Can upload a small subset of the data that shows the same problem?

But I realized we may not need to see the data at all. If any of your xi or yi data are negative, then gamfit will give that error, because a gamma distribution cannot generate negative values.

[Arly Getha Purba]

but there is no negative values both for the historical and the observation data. Am I have to change the distribution?

[Jeff Miller]

Any zeros? Does this work:

Pxi = gamfit(xi(xi>0));