Travail de recherche/Working paper
| Résumé : | High-dimensional time series may well be the most common type of dataset in the socalled“big data” revolution, and have entered current practice in many areas, includingmeteorology, genomics, chemometrics, connectomics, complex physics simulations, biologicaland environmental research, finance and econometrics. The analysis of such datasetsposes significant challenges, both from a statistical as from a numerical point of view. Themost successful procedures so far have been based on dimension reduction techniques and,more particularly, on high-dimensional factor models. Those models have been developed,essentially, within time series econometrics, and deserve being better known in other areas.In this paper, we provide an original time-domain presentation of the methodologicalfoundations of those models (dynamic factor models usually are described via a spectralapproach), contrasting such concepts as commonality and idiosyncrasy, factors and commonshocks, dynamic and static principal components. That time-domain approach emphasizesthe fact that, contrary to the static factor models favored by practitioners, the so-called generaldynamic factor model essentially does not impose any constraints on the data-generatingprocess, but follows from a general representation result. |
