[Modelling of temporal data and the role of graphics]

The importance of graphics for temporal data modelling, which is often used in ecopathology, is illustrated through 2 examples. The first concerns a univariate time series (tank milk germ measurements). It uses correlograms (choice of the model), plots of residuals of the model (diagnostics) and principal component analysis graphical outputs (comparison of time series). The second concerns climate multivariate monthly time series. A 3-way analysis and graphical outputs are used to give a mark to each station, describing its climate for a set of months.