Schedule May 24, 2018
Defining a Trend in A Time Series: Can We Tell How Warm it's Getting?
Juan Restrepo (OSU)

A useful machine-learning application would be to find an "executive summary" of a time series. By this I mean another time series, which when subtracted from the original series yields a series that can be captured by a simple statistical model (e.g. a normal variate). The utility of such a "tendency" is that it would encapsulate the most critical aspects a dynamicist might need to explain if this signal has a rational basis. The intrinsic time decomposition is shown to be a diffusion filter that can handle multiscale dynamics. A "tendency" can be produced from an intrinsic time decomposition, followed by a selection criterion that favors empirical statistical symmetries and information content. The tendency, when applied to a climate signal, will show the warming trend as well as the preponderance of the ocean in controlling the time scales of fluctuations.

To download: Right-click and choose "Save Link As..."   (Other video options)

To begin viewing slides, click on the first slide below. (Or, view as pdf.)

[01] [02] [03] [04] [05] [06] [07] [08] [09] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]

Author entry (protected)