Robust smoothed canonical correlation analysis for functional data

BOENTE, GRACIELA; KUDRASZOW, NADIA L.

STATISTICA SINICA

STATISTICA SINICA

Año: 2022 vol. 32 p. 1269 - 1293

1017-0405

This paper provides robust estimators for the first canonical correlation anddirections of random elements on Hilbert separable spaces by using robust association andscale measures combined with basis expansion and/or penalizations as a regularizationtool. Under regularity conditions, the resulting estimators are consistent. The finitesample performance of our proposal is illustrated through a simulation study that showsthat, as expected, the robust method outperforms the existing classical procedure whenthe data are contaminated. A real data example is also presented.