Upgrade time-varying models using chunks and specifying prior parameters

[ntorresd]

• Please check if the PR fulfills these requirements

• [x] I have read the CONTRIBUTING guidelines

• [ ] A new item has been added to NEWS.md

• [x] Tests for the changes have been added (for bug fixes / features)

• [x] Docs have been added / updated (for bug fixes / features)

• [x] Checks have been run locally and pass

• What kind of change does this PR introduce? (Bug fix, feature, docs update, ...) This PR:

  • Upgrades the time-varying models to allow for force-of-infections estimations by chunks rather than year by year.
  • Updates testing strategy using chunks and data simulation
  • Partially solves #118

• What is the current behavior? (You can also link to an open issue here)

Currently, one value of the force-of-infection is estimated for every year in the time spanned by the serosurvey.

data(chagas2012)
serodata <- prepare_serodata(chagas2012)
seromodel <- fit_seromodel(serodata, foi_model = "tv_normal")

• What is the new behavior (if this is a feature change)? Now chunks can be specified either by size:

  seromodel <- fit_seromodel(serodata, foi_model = "tv_normal", chunk_size = 10)

  

or explicitly specifying the chunk structure:

chunks <- rep(c(1, 2, 3, 4, 5), c(10, 10, 15, 15, max(serodata$age_mean_f)-50))
seromodel <- fit_seromodel(serodata, foi_model = "tv_normal", chunks = chunks)

• Does this PR introduce a breaking change? (What changes might users need to make in their application due to this PR?) Yes, two breaking changes are introduced in this PR:
  1. The use of chunks is a breaking change allowing users to specify prior knowledge about the force-of-infection by assigning constant values of the force-of-infection to specific time intervals. The default behavior is chunk_size=1, which recovers previous results.
  2. Now the user can specify the initial location and scale parameters of the distributions used in the stan files by means of foi_location and foi_scale parameters in fit_seromodel. The interpretation of these parameters may vary depending on the specified model. To recover previous results, users should use:

     
     # tv_normal model
     fit_seromodel(
     serodata = serodata,
     foi_model = "tv_normal",
     foi_location = 0,
     foi_scale = 1,
     ...
     )

tv_normal_log model

fit_seromodel( serodata = serodata, foi_model = "tv_normal", foi_location = -6, foi_scale = 4, ... )

* **Other information**: