Data-Driven I/O Structure Learning With Contemporaneous Causality

摘要

In the era of big data, industry and public policy are able to make use of large amounts of data for policy decisions. The proliferation of cheap sensors and fast communication enables policy makers to consider complex networks as a whole, using time series data from many sources to model the system. The input/output structures of such systems are helpful in understanding how they work and designing new control laws. This article introduces the causal dynamic graph (CDG) model, which defines this structure explicitly. We provide a data-driven method for recovering the input/output structure of a CDG when every process is measured. We then discuss some of the implications of incomplete measurements on the graphical modeling and structural identification problem; we show that many relevant cases are equivalent to the simpler case where sensors are either perfect or completely missing. This will make the problem of graphically modeling such systems more tractable.