Time Series Data with Semi-Reflective Boundaries: Single-Pass and Iterative Methods

• Activity Number: 482
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Jeffrey D. Dawson*+ and Amy M. Johnson
• Companies: The University of Iowa and Univ of Iowa
• Keywords: Spatial/temporal model ; Grid search ; Driving studies ; Likelihood

Abstract

Some time series data have values that tend to reflect back towards the middle of their range. For example, a motor vehicle that that starts to cross its lane boundaries may be promptly directed back toward the middle of the desired driving lane. Dawson et al (2010) proposed a model to accommodate such boundaries, using third-order polynomial projections plus an error term whose sign and magnitude are modeled stochastically. They fit this model using a single pass approach involving a combination of linear and logistic regression. In this report, we explore iterative likelihood-based methods to fit this model, as well as a new single pass method. We do these comparisons using simulated data, as well as vehicular control data from 127 middle-aged and elderly drivers. This work was supported by NIH/NIA awards AG17177 and AG15071.