Classification of the Aneurisk65 dataset using PCA for partially observed functional data

When functional data are observed over a domain that is subject-specific, most of the techniques for functional data analysis are invalidated. Recently, new methods able to handle this situation were developed and in particular we focus on well-known functional PCA. With the aim of classifying the Aneurisk65 dataset, we apply a few possible methods and we show that carrying out the analysis over the full domain, where at least one of the functional data is observed, may not be the optimal choice. This is also confirmed in a simulation study, where the best interval for classification lies between the common domain and the full domain.