1: \begin{abstract}

2: 		We present a forward sufficient dimension reduction method for categorical or ordinal responses by extending the outer product of gradients and minimum average variance estimator to categorical and ordinal-categorical generalized linear models. 

3: 		Previous work in this direction extend forward regression to binary responses, and are applied in a pairwise manner for multi-label data, which is less efficient than our approach. 

4: 		Like other forward regression-based sufficient dimension reduction methods, our approach avoids the relatively stringent distributional requirements necessary for inverse regression alternatives. 

5: 		We show consistency of our proposed estimator and derive its convergence rate. 

6: 		We develop an algorithm for our methods based on repeated applications of available algorithms for forward regression. 

7: 		We also propose a clustering-based tuning procedure to estimate the tuning parameters. 

8: 		The effectiveness of our estimator and related algorithms is demonstrated via simulations and applications.          

9: 	\end{abstract}

10: