An in-development R package and a Bayesian hierarchical model jointly fitting multiple "local" wastewater data streams and "global" case count data to produce nowcasts and forecasts of both observations
The model expects no NA values in the wastewater data, but its possible (likely) that when a measurement is below the LOD, it will be NA. Since these are censored and the model treats them as such, we don't want to just remove them. Instead, we need to be really explicit that the model expects only numeric values for concentrations and give directions to the user as to how to pre-process their data to replace observations below the LOD with a value that is below the LOD.
Requirements
[ ] If there are NA values in input data, ask users to explicitly indicate a value less than the LOD when values are below the LOD
[ ] Ask users to report data on the log scale, handle that accordingly
Context
@akeyel passed in NAs for his values that were below the LOD, and when it errored assumed they should be left out, when in fact these actually should be included as they will be treated as censored
Problem
The model expects no NA values in the wastewater data, but its possible (likely) that when a measurement is below the LOD, it will be NA. Since these are censored and the model treats them as such, we don't want to just remove them. Instead, we need to be really explicit that the model expects only numeric values for concentrations and give directions to the user as to how to pre-process their data to replace observations below the LOD with a value that is below the LOD.
Requirements
Context