Grouped variable importance with random forests and application to multiple functional data analysis

In this paper, we study the selection of grouped variables using the random forests algorithm. We first propose a new importance measure adapted for groups of variables. Theoretical insights of this criterion are given for additive regression models. The second contribution of this paper is an original method for selecting functional variables based on the grouped variable importance measure. Using a wavelet basis, we propose to regroup all of the wavelet coefficients for a given functional variable and use a wrapper selection algorithm with these groups. Various other groupings which take advantage of the frequency and time localisation of the wavelet basis are proposed. An extensive simulation study is performed to illustrate the use of the grouped importance measure in this context. The method is applied to a real life problem coming from aviation safety.