Multiscale analysis for time series correlation networks


Daniel Rockmore

Dartmouth College

Northeastern University

Thursday, April 10, 2008


Talk at 4:30 p.m. in 509 Lake Hall

Tea at 4:00 p.m. in 544 Nightingale Hall


Abstract:   The analysis of network structure is an important part of the study of complex systems. In this talk we focus on the analysis of networks defined by the correlation structure of time series. Examples include neuronal activity, communications networks, and financial networks. We describe a new method for exposing multiscale structure in such correlation networks, that is different from the various hierarchical clustering techniques that are often applied in this setting. Our chief example is the time series defined by the close prices of the equities market for which the methodology reveals some nontrivial and interpretable homology in the network. This is joint work with Greg Leibon, Scott Pauls, and Robert Savell.

Here are some directions to Northeastern University. Lake Hall and Nightingale Hall can be best accessed from the entrance on the corner of Greenleaf Street and Leon Street. The two halls are connected, with no well-defined boundary in between. In particular, 509 Lake Hall is on the same corridor as 544 Nightingale Hall.

There is free parking available for people coming to the Colloquium at Northeastern's visitor parking (Rennaisance Garage). The entrance is from Columbus Avenue. If coming by car, you should park there and take the parking talon. After the lecture, you may pick up the payment coupon from Andrei Zelevinsky.

Home Web page:  Alexandru I. Suciu  Comments to:  
Posted:: April 2, 2008    URL: