Network Temporalization

Overview

The Network Temporalization tool unfolds a Bayesian network over time so it can represent a time series.

Temporalization Options

The Temporalization tool provides the following options.

You can specify a temporal step size. This is particularly helpful when a certain lag structure in the data-generating process is already known. For instance, if the data was recorded with a daily frequency and you know of a weekly seasonal pattern, you might use a step size of 7.

Row identifiers can be used to distinguish between time series, for example of different subjects, that are stacked in a single column.

The variables and classes are named from “t-n” to “t”, following the convention used in econometrics.