distributionFitter: Lognormal distribution

Hi

I wanted to fit my Data to the lognormal distribution, but I get this error message : "The data in X must be positive".

I am pretty sure that the data and the frequency arrays are positive (>=0).

How can I ovecome this issue.

[Torsten]

Maybe there are NaN, Inf or -Inf values in your data vector ?

What does

any(Data(1,:) < 0)

return ?

Check it. We cannot because you didn't include the data.

[Mohamed Zied]

>> any(Data(1,:) < 0)

ans =

logical

0

I did, none of the values is negative. Actually, Data represents the bin edges. So logically, it can be negative. This error pops up only with Lognormal distribution.

[Torsten]

So you don't use the Data(1,:) directly to fit the lognormal distribution, but some preprocessed data that you don't know exactly ?

What about

any(Data(1,:) <= 0)

?

[Mohamed Zied]

Yes, I found some zeros.

I have to remove these values to be able to fitt my data to the Lognormal distribution.

Thank you for your help

[Mohamed Zied]

I predefined the bin edges. for some bins the count is 0. Thus, the error.

Answer 1

[Steven Lord]

I am pretty sure that the data and the frequency arrays are positive (>=0).

0 is not positive. What does this command show?

any(Data(1, :) == 0)

[Mohamed Zied]

You are right. There are some zeros in my Data. I have to remove these values to be able to fitt my data to the Lognormal distribution.

Thank you for your help