Hi all.
I am using dlnm package in R, in order to assess temperature-mortality association in a time series analysis. I have daily values of temperature and deaths for 11 years.
I want to perform pacf (Partial Autocorrelation Function) in order to decide which exposure variable (temperature or apparent temperature) is best for my dataset.
Are there any ideas on how to achieve that?
Google search pacf function in R. This should do it
Best wishes David Booth
David Eugene Booth thank you for your answer. I have googled pacf but I cannot understand in
Pacf( x, lag.max = NULL, plot = TRUE, na.action = na.contiguous, demean = TRUE, ... )
what x should be since I want to check my 2 models and decide which is best, based on their pacf values...
Ah I see you probably want cross corelation not pacf. See Shumway Time series book in the z-library for download.. or cross validation in that case look at Steyerburg Clinical prediction models good luck with your research, , David Booth
The residuals that you obtain from your model are the "x" input in pacf(), and any model that works with it without a DLNMs should also work with a DLNM included. Put briefly, it shouldn't be any more difficult to use pacf with a DLNM.
A simple way to code it is:
model
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size