LIP External Colloquium: Steve Boker, University of Virginia, USA - "Parameter Selection for Windowed Cross Correlation"
- Date: Jun 10, 2026
- Time: 01:00 PM - 02:00 PM (Local Time Germany)
- Speaker: Steve Boker, University of Virginia, USA
- Location: Max Planck Institute for Human Development, Lentzeallee 94, 14195 Berlin
- Room: 299
- Host: LIP
- Contact: seklindenberger@mpib-berlin.mpg.de
- Topic: Lectures
Steve Boker, University of Virginia, USA - "Title TBA" will present:
„Parameter Selection for Windowed Cross Correlation“
Abstract: Windowed Cross Correlation (WCC) has long been in use for estimating the relationship between nonstationary time series from pairs of sources such as human movement, heart rates, EEG, and BOLD. But WCC requires selection of parameters for window size, maximum time lag, and time resolution for each of these. Every type of time series has different requirements for these parameters, making the use of WCC difficult. A new R package called wcc is now on CRAN and can be used to estimate the power to distinguish between real and surrogate time series for a variety of aggregation statistics. One may use wcc to perform a grid search on a holdout set from the time series of interest in order to select appropriate parameters for WCC. This talk provides an introduction to why one would want to use WCC as a method and the particulars of using the wcc R package. Upcoming features will be discussed that are intended to distinguish onsets and offsets of intervals of approximate symmetry (sometimes referred to as synchrony) between paired time series.
Join online:
https://mpib-berlin.webex.com/mpib-berlin/j.php?MTID=mf55ba588b527ad452de24d6815027887
Meeting number: 2744 844 6109
Password: 8bKvH8EbCe6
„Parameter Selection for Windowed Cross Correlation“
Abstract: Windowed Cross Correlation (WCC) has long been in use for estimating the relationship between nonstationary time series from pairs of sources such as human movement, heart rates, EEG, and BOLD. But WCC requires selection of parameters for window size, maximum time lag, and time resolution for each of these. Every type of time series has different requirements for these parameters, making the use of WCC difficult. A new R package called wcc is now on CRAN and can be used to estimate the power to distinguish between real and surrogate time series for a variety of aggregation statistics. One may use wcc to perform a grid search on a holdout set from the time series of interest in order to select appropriate parameters for WCC. This talk provides an introduction to why one would want to use WCC as a method and the particulars of using the wcc R package. Upcoming features will be discussed that are intended to distinguish onsets and offsets of intervals of approximate symmetry (sometimes referred to as synchrony) between paired time series.
Join online:
https://mpib-berlin.webex.com/mpib-berlin/j.php?MTID=mf55ba588b527ad452de24d6815027887
Meeting number: 2744 844 6109
Password: 8bKvH8EbCe6