0

I am looking for some debugging help on the function code below. A little context. I have run a mixed logit model using the mlogit package in R, and have generated a panel model without correlations between the random coefficients.

In total, I have 11 coefficients that I am looking to use to generate a predictive model by running a simulation. I am aiming to achieve this by running a function, noted below (a variation on the code in the book Simulations in R for Marketing Research and Analytics:

predict.mnl <- function(model, data, nresp=2000) 
  { 
  library(MASS)
  data.model <- model.matrix(update(model$formula, 0 ~ .), data = data)
  coef.mu <- model$coef[1:11] **#Coefficients**
  coef.sd <- abs(model$coef[12:22]) #**SD to end, corrected for abs value**
  for (j in 1:11) { 
    draws[j,] <- rnorm(n=nresp, coef.mu[j], coef.sd[j])
    shares <- matrix(NA, nrow=nresp, ncol=nrow(data)) 
    }
  for (i in 1:nresp) {
    utility <- data.model%*%draws[i, ]
    share = exp(utility)/sum(exp(utility))
    shares[i, ] <- share
  } 
  
  cbind(colMeans(shares), data)
  }

predict.mnl(m3, new.data)

Effectively what I am trying to do is take the coefficients and standard deviations, generate individual level estimates of each of the coefficients for 2000 'participants' in a new matrix by drawing random samples (in the for (j in 1:11) loop), create the predicted share for each (in the (i in 1:nresp) loop) and multiply each by a model by calling the function for (i in 1:nresp.

However, I am getting an error

Error in draws[j, ] <- rnorm(n = nresp, coef.mu[j], coef.sd[j]) : 
  incorrect number of subscripts on matrix

I presume if is something to do with the for loop, but I am a complete n00b so have no idea how to debug.

Please help!

Thanks in advance community!

Rob M
  • 1
  • Any chance that you can share some of the data you're using when you execute this function? It looks like you're fairly new to SO; welcome to the community! If you want great answers quickly, it’s best to make your question reproducible. This includes sample data (e.g., data.frame(x=...,y=...), like the output from dput(head(dataObject))) and what type of output you are expecting. Check it out: [making R reproducible questions](https://stackoverflow.com/q/5963269). – Kat Jan 17 '22 at 04:57

0 Answers0