R: ksvm function - error: object is not a matrix

Question

I seem to be encountering an error using the ksvm function. Each time I try to run the below code I get an error message

Error in model.frame.default(data = ..1, formula = x) : 
object is not a matrix

I have asked colleagues, but no one seems able to replicate this (for whatever reason). Any assistance would be appreciated. Please see the code below:

Note: I have already tried using as.matrix. I get the same error

library(kernlab)

#Import airquality df
aq_df<-airquality

#Function to replace NA's
replaceNAs<-function(df,df_col)
{
  for (i in 1:dim(df)[1])
  {
    if(is.na(df_col[i]))
    {
      #print(df_col[i])
      df_col[i]<-mean(na.omit(df_col))
    }
    
  }
  return(df_col)
}

#Call the function to replace NA's with mean
aq_df$Ozone<-replaceNAs(aq_df,aq_df$Ozone)
aq_df$Solar.R<-replaceNAs(aq_df,aq_df$Solar.R)
aq_df$Wind<-replaceNAs(aq_df,aq_df$Wind)
aq_df$Temp<-replaceNAs(aq_df,aq_df$Temp)

#Setup ksvm
randIndex<-sample(1:dim(aq_df)[1])
cutPoint_23<-floor(2*dim(aq_df)[1]/3)


trainData<-aq_df[randIndex[1:cutPoint_23],]
testData<-aq_df[randIndex[(cutPoint_23+1):dim(aq_df)[1]],]

svmOutput<-ksvm(type ~.,data=trainData, kernel="rbfdot",kpar="automatic",C=5,cross=3,prob.model=TRUE)

Answer 1

The variable type isn't in the data frame. If you had a variable called type , it would work just fine:

## add a variable named type
aq_df$type <- sample(1:2, nrow(aq_df), replace=TRUE)

## proceed as above
#Setup ksvm
randIndex<-sample(1:dim(aq_df)[1])
cutPoint_23<-floor(2*dim(aq_df)[1]/3)


trainData<-aq_df[randIndex[1:cutPoint_23],]
testData<-aq_df[randIndex[(cutPoint_23+1):dim(aq_df)[1]],]

svmOutput<-ksvm(type ~.,data=trainData, 
  kernel="rbfdot",kpar="automatic",
  C=5,cross=3,prob.model=TRUE)

svmOutput
# Support Vector Machine object of class "ksvm" 
# 
# SV type: eps-svr  (regression) 
# parameter : epsilon = 0.1  cost C = 5 
# 
# Gaussian Radial Basis kernel function. 
# Hyperparameter : sigma =  0.161753619421595 
# 
# Number of Support Vectors : 93 
# 
# Objective Function Value : -225.2614 
# Training error : 0.413681 
# Cross validation error : 0.373122 
# Laplace distr. width : 0.952921