scottyaz/coarsedatatools

an R package for doubly interval censored survival data

coarsedatatools's Issues

Add user to use any priors in dic.fit.mcmc

At the moment the priors are hard coded. Add the priors as inputs (with sensible defaults).

Erlang Distribution

Add option for erlang distribution. First for MCMC and then maybe figure out a way with optim.

change dist labels

At the moment, we use the tags "G" for gamma, "W" for Weibull etc. SHould probably change so that we use the full name (i.e. "Gamma").

dic.getSE() does not get SEs for bootstrap

It returns a matrix of bootstrap estimates. Should probably put this in another function since it really doesn't return what the function name suggests.

optim options

looks like we have the option lower used in many of the optim() statements which isn't compatible with all optim methods. We should try to optimize on a transformed scale when possible and get rid of lower (and upper) options.

clarify sdlog vs dispersion for lognormal models

Seems like, for log-normal models, we either need to very explicitly say somewhere that exp(sdlog) = dispersion, or keep dispersion in the output.

We could have a note print out at the bottom of the output for lognormal models only (by modifying the "show" method for our S4 objects).

asymptotic results for quantiles of other distributions

at the moment we only have them for log-normal

plot() for lnorm fits with no bootstrap

Using the code from:
https://gist.github.com/scottyaz/e46731f183b3ec3506f2
when I run,

lnorm.est.1 <- dic.fit(dat.lnorm)
plot(lnorm.est.1)

the resulting plot is not correct. Perhaps an error with the log transformation of the dispersion? The plot window that appears has the same xlim as when I plot(lnorm.est.2) and .3 but the distribution is much flatter.