cfMake and cfChange should be able to handle multi-equation models

[chrisadolph]

The simplest way, I think, would be to allow for vectors of dataframes and/or vectors of formulas in cf objects. For example, if we wanted a zero-inflated model simulator, we would only need one dataframe, but we need two formulas, one from each stage. But for a hierarchical simulator, we would want two data frames and two formulas. Although we don't have support for these models yet, we'll need this capacity for future extensions.

[aserlich]

Hi Chris -- I met with Carolina this afternoon and we had a productive session. I think I have done this for the hurdle models but not yet for multilevel models. Could you have a look at my approach and make sure I am doing this in a sensible way. Basically, the cfMake needs to distinguish between three types of options. The code is here: https://github.com/aserlich/tile-simcf/blob/master/simcf/R/cfset.R

This code shows how it minimally works...

data <- UScrime
data$Prob[1] <- 0 #add a logit bounded variable
model <- (y ~ log(M) + So + log(Ed) + log(Po1) + log(Po2)
              + log(LF) + log(M.F) + log(Pop) + log(NW) +log(U1)
              + log(U2) + log(GDP) + log(Ineq) + logitBound(Prob) +
              log(Time))

model1 <- (y ~ log(M) + So)
model2 <- (y ~ logitBound(Prob) + So)

znb <- list(model1, model2)
cfMake(formula = model, data = data)
cfMake(formula = znb , data = data)

[aserlich]

CfMake has a “hull” option. However, it is currently not in the helpful and doesn’t appear to be working properly. Could you tells us what if anything you want done with that.

[aserlich]

Hi Chris,

I have now implemented hierarchical models in cfMake. Before I change cfSet, I want to make sure I am on the right track. Are you getting these messages?

testfile is: https://github.com/aserlich/tile-simcf/blob/master/testcfMake.r code file is: https://github.com/aserlich/tile-simcf/blob/master/simcf/R/cfset.R