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.

Play Flash full motion video, or Flash lower bandwidth video. (Or, right-click to download the 3gp file.)

Play QuickTime full motion movie [ or Stream | or Download ]
Or play QuickTime lower bandwidth slideshow [ or Download ] Or [ Download the Podcast ].

Begin streaming RealMedia. (Or, right-click to download the audio file.)

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)