Logistic regression in matlab using mnrfit

Question

I'm trying to use the mnrfit function but I get the error

If Y is a column vector, it must contain positive integer category numbers. .

My data is in a double and my Y values are floats , eg 0.6667. Is there a way that I can adjust my data to be able to use the mnrfit function?

Thanks in advance! An unexperienced beginner

Answer 1

Y should be a "nominal outcome", ie non-continuous, to use mnrfit . We don't need to turn Y into integers, just categoricals. A categorical array is discrete as far as MATLAB is concerned, regardless whether the categories are represented by double values.

X = rand(5,3); % Predictors (should be double or single)
Y = rand(5,1); % Response (doubles, will cause error)

B = mnrfit( X, Y )
% ERROR: If Y is a column vector, it must contain positive integer category numbers. 

B = mnrfit( X, categorical(Y) )
% No error, regression matrix B is output successfully.

Be careful , if you're expecting a continuous response variable (hence why Y is a vector of doubles) then mnrfit may not be suitable in the first place!

---

Note the valid data types are specified in the docs

Y can be one of the following:

• An n-by-k matrix, where Y(i,j) is the number of outcomes of the multinomial category j for the predictor combinations given by X(i,:). In this case, the number of observations are made at each predictor combination.

• An n-by-1 column vector of scalar integers from 1 to k indicating the value of the response for each observation. In this case, all sample sizes are 1.

• An n-by-1 categorical array indicating the nominal or ordinal value of the response for each observation. In this case, all sample sizes are 1.

Answer 2

I have quick question when I use "B = merit( X, categorical(Y) )" I am getting a negative Beta value but the actual Beta value is positive of what I got. EX) actual Beta = 50 , -90 , 8 what I am getting= -50, 90 ,-8