B1896
Title: A functional logistic regression model with non-independent functional variables
Authors: Cristhian Leonardo Urbano Leon - Universidad de Granada (Spain) [presenting]
Manuel Escabias - University of Granada (Spain)
Ana Maria Aguilera - University of Granada (Spain)
Abstract: A proposal is presented to extend the functional logistic regression models, which model a binary scalar response variable from a functional predictor to the case when the functional variables are not independent but paired, i.e., the same functional variable under different experimental conditions. We assume that the curves of the functional predictor and the parameter function of the model belong to the same finite-dimensional subspace of the space $L_{2}$ of square-integrable functions over the same closed real interval. This approach allows the use of basis expansion methods for the treatment of functional data. The proposal is contextualized with an application to biomechanical functional data that records the position of joints (angles they form with certain axes) with respect to the total walking cycle.
