CODEWITHSUNDEEP
r
user4745212
user4745212

Reputation: 51

R Neural Networks - Iris Dataset Confusion Matrix

I don't how to figure the following: 1. Which 30 elements did it predicted ? 2. I can't get the ConfusionMatrix working.

Any help appreciated. Thank you.

library(nnet)
attach(iris)

library(caret)
set.seed(3456)
trainIndex <- createDataPartition(iris$Species, p = .8,
                               list = F,
                               times = 1)

irisTrain <- iris[ trainIndex,]
irisTest  <- iris[-trainIndex,]

irispred <- nnet(Species ~ ., data=irisTrain, size=10)
predicted <- predict(irispred,irisTest,type="class")

output:

predicted <- predict(irispred,irisTest,type="class")

predicted [1] "setosa" "setosa" "setosa" "setosa" "setosa" "setosa" "setosa"
[8] "setosa" "setosa" "setosa" "versicolor" "versicolor" "versicolor" "versicolor" [15] "versicolor" "versicolor" "virginica" "versicolor" "versicolor" "versicolor" "virginica" [22] "virginica" "virginica" "virginica" "virginica" "virginica" "virginica" "virginica" [29] "virginica" "virginica"

Confusion Matrix errors, not sure what the should the 2nd argument be:

confusionMatrix(predicted, iris$Species) Error in table(data, reference, dnn = dnn, ...) : all arguments must have the same length

confusionMatrix(predicted, irisTest, positive=1) Error in sort.list(y) : 'x' must be atomic for 'sort.list' Have you called 'sort' on a list? confusionMatrix(predicted, iris, positive=1) Error in sort.list(y) : 'x' must be atomic for 'sort.list' Have you called 'sort' on a list?

Upvotes: 1

Views: 2886

Answers (1)

thie1e
thie1e

Reputation: 3688

The second argument should be the reference classes which are in irisTest$Species. irisTest is the complete test data including all the other columns and iris$Species are the species for the whole data, not just the test set.

> confusionMatrix(data = predicted, reference = irisTest$Species)
Confusion Matrix and Statistics

            Reference
Prediction   setosa versicolor virginica
  setosa         10          0         0
  versicolor      0          9         0
  virginica       0          1        10

Overall Statistics

               Accuracy : 0.9667           

[...]

Upvotes: 1

Related Questions