0

When running mclust on my test dataset, I get the following error:

Error in cdensEEV(data = data, logarithm = TRUE, parameters = parameters, : NA/NaN/Inf in foreign function call (arg 1) In addition: Warning message: In cdensEEV(data = data, logarithm = TRUE, parameters = parameters, : NAs introduced by coercion

My code is below:

library(ISLR)
attach(Auto)
library(mclust)
library(MASS)
library(class)
attach(Weekly)
Auto$mpg01 <- ifelse(Auto$mpg > median(Auto$mpg),1,0) 
odd <- seq(from = 1, to = nrow(Auto), by = 2)
even <- seq(from = 2, to = nrow(Auto), by = 2) 
X.train <- Auto[odd, -10] 
Class.train <- Auto[odd, 10] 
X.test <- Auto[even, -10] 
Class.test <- Auto[even, 10] 
AutoMclustDA <- MclustDA(X.train, Class.train) 
summary(AutoMclustDA, parameters = TRUE)
summary(AutoMclustDA, newdata = X.test, newclass = Class.test)

When running the initial summary, everything works fine, but when I run the second summary, and specifically the newclass = Class.test portion, I get that error. I don't have any NA values in the dataset, and when I did it previously for the Weekly dataset, I had no issues.

Can anyone help? Thank you in advance.

Adam

Bobo The Fool
  • 15
  • 5

1 Answers1

0

So, not so much an answer as a work-around. One of the variables in the dataset is the name of the vehicle, so when I removed that variable, which really has no bearing on the analysis, and adjusted the train and test notations accordingly (ex. X.train <- Auto[,-9]), the summary worked, both for MclustDA as in the above code and also when you add in the modelType = "EDDA".

Bobo The Fool
  • 15
  • 5