Dear Researchers,
I am doing timeseries modelling on 154 observations and my number of Independents is 21. I am using three years of weekly data.
And for the analysis I am using "Bayesian Dynamic Linear Modeling Using Kalman Filter" where i get the beta coefficient for each time period
So it is fine to do the modelling with this number of observation or do I need to reduce the independents.
Thanks in Advance!
Danish
In Bayesian modelling you update your knowledge. Even if you only had one observation you could (in principle) get a better answer than you already had. In practice you are not very interested in such a situation as you will learn very little. But sometimes you will have to.
If you don't understand this part of Bayesian analysis, I advise you to first take a good book on Bayesian Inference and study it toroughly. It is the core of Bayes rule.
For 154 observations, I think 21 variables are too many. I think you should reduce the number of variables. Regards
I agree: 21 variables is too many. Take a closer look at your model to see if you can't reduce this number
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Control Theory
- Control Systems