What time series type analysis in R should I use?

Question

I'm trying to determine what type of analysis I should use in R for my data/question. I'm thinking some type of time lag or time series analysis?

My question is how the duration of a behavior changes, on average, for a number of different individuals, in the 5 minutes before versus 5 minutes after an event occurs. Basically, I want to know if there is a change in the duration of a behavior that is contingent on the event occurring.

It needs to be automated because I have a LOT of data (multiple days of data for many different individuals). I have the time that an event occurs and the time (and duration) of a behavior I'm interested in (HH:MM:SS). I want to see the average change in the duration of a behavior per individual for 5 minutes before the event occurs and 5 minutes after the event occurs.

Does anyone have a recommendation for what type of analysis would be best for this question?

Answer 1

I am really not an expert on this, but it seems to me that panel regression might help, since there are both time series as well as individual level variation.

For the 5 minute before/after analysis, I am thinking maybe introducing dummy varaibles (DV) could be helpful? Ie. DV1 could be 0 or 1 if the minutes part of your time stamp is <=5 minutes from the event timestamp and similarly DV2 for >=

Answer 2

You might want to look into behavioral change point analysis. The bcpa package in R can help you.

see
http://cran.r-project.org/web/packages/bcpa/vignettes/bcpa.pdf

and http://wiki.cbr.washington.edu/qerm/index.php/Behavioral_Change_Point_Analysis