0

I want to estimate a multinomial logit model(mixed logit) with heteroskedastic customer "brand preferences" (random parameter logit model) using mlogit() function.

The dataset is the saltine Cracker dataset and users can choose among 4 brands. Additionally, there are 3 product specifics: price, feature, display.

However, I get this error: "Unknown random parameter" when I want to assign my brand intercepts (heterogeneity for brand preferences) to rpar.

Here is my code:

library(mlogit)
data(Cracker, package ="mlogit")
Cracker = subset(Cracker, id <= 100)
cracker.data <- mlogit.data(Cracker, choice = "choice", shape = "wide", varying = 2:13, id = "id")
result = mlogit(choice ~ disp + feat + price, 
                       data = cracker.data, 
                       rpar = c("sunshine:(intercept)" = "n", 
                                "keebler:(intercept)" = "n", 
                                "nabisco:(intercept)" = "n"), 
                       reflevel = "private", panel = TRUE)

Thank you in advance for your time and help.

mahsa paridar
  • 21
  • 4

1 Answers1

1

I realized that the brand intercept should be written exactly as it is in the summary coefficient of standard homogenous multinomial logit. I should have written: "(Intercept):sunshine etc."

mahsa paridar
  • 21
  • 4