What to do if the model coefficients are complex?

Hi everyone! I am trying to develop a mupltiple regression. In particular, I got 3 categories: trochateric fracture, neck fracture and control patients. I am using teh following data:

• training_data: 300x13 double (13 columns because I added a column of ones, as requested by the function I use)
• classes_training: 300x1 (where the class can be 1,2 or 3)
• test_data: 51x12
• classes_test: 51x1

I have normalized both training_data and test_data. Now, I apply the functions:

MR = mnrfit( classes_training, training_data);

eval = mnrval(MR ,test_data);

But the following error is displayed:

"Error using mnrval (line 189) The sizes of B and X are incompatible. "

Actually the sizes are correct, because MR is 2x12 and test_data is 51x12. But, I don't know why, MR is composed by complex coefficients. I suppose this is the problem, but I don't know why and how to solve it. Could you help me? Thank you in advice!