After considerable search and thinking session, (and of course, conversation with Mike Young always help), it become increasingly clear to me that survival analysis may be even more appropriate than data mining. In some circle, data mining may be construde as fishing for data, and is always suspect. A more direct data analysis method would of course be desirable, but which one?
I have already come across various time-stamping methods, and the most closely “sounding” one may be time-series analysis. However, after considerable search into the method, it turned out to be closer to financial analysis (as in moving averages) than what I am hoping for. No go…
Then, I came across some literature on time-to-event analysis… (as in this BMJ article). I think I have found the right thing. (Mike seemed to think so, too.)
Information Trails is an innovative user behavior analysis framework for in situ assessment within multi-user virtual environments, including serious games, virtual worlds and simulations.
Design by
Subscribe to entries
Subscribe to comments
All content © 2010 by Information Trails