Master-project

The report is included as the file Master_project_Halvard - Finished.pdf and the figures and plots are created with the Rmd files in this repository. The code is briefly described below.

Master project - AGMRF.Rmd

Firstly, this file looks at first and second order random walks in time and implements them and make figures. Secondly, there are multiple regeneric definitions both for a RW1 and an adaptive RW1. They also both have definitions where the precision matrix $Q$ is defined inside the rgenric or supplied as an argument. Then follows data generation for all the mean trends and some testing and examples with INLA and the different latent models

Wakefield reprodction for comparison.Rmd

The first half of the file concerns the reproduction of the results in the Wakefield article. Currently, it assumes that the data generation from the file above have been run and is in the global memory, but they can also be saved and loaded. Then the model fitting, evaluation and plotting is done. The prior_str is defined at the top, and is either set to PC, Gamma0,00005 or Gamma0,005 for the respective priors. The second half of the file is similar code for the harmonious mean structures, it uses the function mod_eval_W from above, and make sure the file location at the bottom of the function is correct, I had two different folders for the results. The rest is slightly modified to fit this example, and very similar to the above code. Note, that we again assign the prior_str for each prior.

Application for Norwegian death rates.Rmd

This file creates all the plots and figures concerning the applied study in the master project. It can be automated with some nested for loops, but for now you manually choose the observational likelihood, either poisson or nbinomial and a prior, which is either $Gamma(1, 0.00005), Gamma(1, 0.005)$ or $PC(1, 0.01)$. You must also manually adjust the title to fit the specific situation.