Beta formulations

• bs: JD formulation σ = ab/(a+b)
• bt: Standard formulation θ = a + b
• Conceptually speaking, if bs and bt look about the same for p ~ 1/2, then we expect bs to have relatively larger "information" (thus narrower distributions) when p is far from 1/2. It would be good to do the calculation

Data sets

Input

• sgtf_ref
• bsfake (simulated using sigma formulation)
• btfake (simulated using theta formulation)

Series types

sr/sg for the data differentiated/not by reinfection

• · ll/agg/ts for linelist/aggregated/aggregated-with-proportions(time series)

Time points

• chop2 (take the last two days off of the data series)

sensitivity/specificity mle2 analysis (needs to be re-combined)

beta binomial paradigm

• bsfit or bsfit

Parameter assumptions

• ssfix – fix logain and lodrop
• ssfitspec – fix logain
• ssfitboth – fix neither

Special names (search forcelink in Makefile)

• main.sx.ts.rds is an alias for sgtf_ref.chop2.sx.ts.rds (our current main data set)

• bsfake and btfake (above) are also special names

Fake data

What_ever_!!