Modeling Change: Incorporating Dynamic Components into Data Analysis
James Ramsay , Ph.D.
University of British Columbia
April 10, 2008
We often need to understand how a system responds to a sudden change in input variables. Is there a time lag before the system begins to change? How quickly does the system adjust to the change? Is this adjustment smooth, or are there transient effects such as initial negative responses or oscillations about the final level before the output settles down? Our own “boy meets girl” memories are a handy source of examples.
These questions are illustrated by a variety of interesting data, including smallpox deaths in Montreal, mudslide conditions in Vancouver, symptom flares for the disease lupus, and analog responses by human subjects to changes in pitch.
A short introduction to new methods for estimating models defined by differential equations or dynamic systems from noisy data will be offered.