A robust partial least squares approach for function-on-function regression

The function-on-function linear regression model in which the response and predictors consist of random curves has become a general framework to investigate relationship between functional predictors. Existing methods estimate parameters may be sensitive outlying observations, common empirical applications. In addition, these severely affected by such leading undesirable estimation prediction results. A robust method, based on iteratively reweighted simple partial least squares, is introduced improve accuracy presence outliers. performance proposed method number squares components used model. Thus, optimum determined via data-driven error criterion. finite-sample investigated several Monte Carlo experiments an data analysis. nonparametric bootstrap applied construct pointwise intervals for function. results are compared with some existing illustrate improvement potentially gained method.