For those who frequently work in markov models, I am interested in knowing the minimum number of time points/observations to get stable estimates of parameters.
I have some follow ups, too:
- Does the number of states determine the minimum number of timepoints/observations?
- Is it different for hidden vs. observed markov models?
I am currently working on a problem that is with observed markov models with only 2 states. I only have 16 time points, which feels low, but I'm not sure.
Any sources would be helpful too. Thanks!
Hi Christopher Loan
Please have a look at these article
Article A Minimal Continuous-Time Markov Pharmacometric Model
Sourangshu Ghosh, thank you for this paper!
I’ve been looking into CTSEM for another project with uneven time intervals. Did not know about CTMM.
See this journal:
On the Use of Mixed Markov Models for Intensive Longitudinal Data
. Multivariate Behavioral Research, Volume 52, 2017 - Issue 6. S. de Haan-Rietdijk, P. Kuppens 📷, C. S. Bergeman, L. B. Sheeber, N. B. Allen & E. L. Hamaker
- https://doi.org/10.1080/00273171.2017.1370364
- Badges
- Science topic
- Similar topics
- Psychology
- Emotion