1: \begin{abstract}

2:  We propose a penalized likelihood method that simultaneously

3: fits the multinomial logistic regression

4: model and combines subsets of the response categories.

5: The penalty is nondifferentiable when pairs of columns in

6: the optimization variable are equal. This encourages

7: pairwise equality of these columns in the estimator, which corresponds to response category

8: combination. We use an alternating direction method of multipliers algorithm

9: to compute the estimator and we discuss the algorithm's convergence.

10: Prediction and model selection are also addressed.

11: 

12: 

13: \par

14: \medskip

15: \noindent \textbf{Key Words:} Multinomial logistic regression; 

16: Fusion penalty; Response category reduction.

17: \end{abstract}

18: