Here we propose a different approach that is viable for any multidimensional case. In this section we focus on the characterization of multidimensional bursts as the very general burst category that includes onedimensional bursts as the simplest case. In the following we omit the features that have been already defined in one-dimensional burst analysis and that are still valid in the general case. Therefore we focus on a set of features, and corresponding quantifying metrics, able to fully characterize multidimensionality. Indeed, multidimensionality features each single burst, not the overall time series, and so models and metrics relative to burst sequence still hold. Here we zoom in the single burst structure to KH7 describe its inner multidimensionality. Nevertheless, these two features fail to describe how often a user switches from one medium to the other. Let us consider, for example text/call bursts. In addition to the one-dimensional burst, where the user decides to perform all activities on a single medium, multidimensional burst is a multiple symbol sequence. Symbols can be more or less interleaved inside a sequence, accounting for how often the user switches between media inside a burst. This way switches divide a burst into sub-sequences, each being a sequence of a sole symbol. An extreme case, very similar to the one-dimensional one, occurs when the burst can be divided in exactly two sub-sequences, one containing text messaging, the other call only. We name this burst a disjoint burst, as the user is definitely separating the two media. Single and disjoint bursts clearly account for a monotone behavior w.r.t selecting a particular type of GW2974 activity; for example, a user may decide to use only one communication medium or to send all texts prior to performing other activities. By contrast, bursts where symbols of different media are interleaved with one another are clearly an observable effect of the multidimensionality and can provide valuable insight into the selection process underlying the user��s activities. For purposes of clarity, we use the term interleaved to identify bursts that are neither disjoint nor one-dimensional. Of course, interleaved bursts exhibit different degrees of interleaving, which account for how often the user changes media or, equivalently, how many subsequences exist in the symbol sequence. In Table 3 we report the comparative analysis between original and shuffled time series, along with the percentage of variation rate.