I have trained a model on a sample size of 2120x10. Now I'm trying to apply the same model to the test data set and I'm having trouble deriving the confusion matrix.
test_predictions <- predict(train_obj, test_data)
test_predictions <- ifelse(test_predictions > 5, 1, 0)
confusionMatrix(as.factor(test_predictions), test_data$outcome, positive="1")
I get an error when calculating the confusion matrix as test_data$outcome has 2135 values. If I use test_data$outcome[1:2120], everything works fine.
Is there a better way to calculate the confusion matrix without restricting the number of values.? Is it correct to restrict the number of values in test_data$outcome?
That doesn't sound right. How can test_data$outcome have 2135 values if test_data only has 2120 rows? Even if there are NAs in test_data's predictors, they will be predicted as NA and then ignored by confusionMatrix.
dat=data.frame(a=rnorm(1000), b=rnorm(1000))
dat=dat %>%
mutate(c=5*(a+b)) %>%
mutate(d=ifelse(c>5, 1, 0))
set.seed(1)
i=sample(1:1000, 750, replace=FALSE)
train_data=dat[i,]
test_data=dat[-i,]
test_data[sample(1:250, 3),1:2]=NA
lr=lm(c ~ a + b, data=train_data)
test_predictions=predict(lr, test_data)
test_predictions=ifelse(test_predictions>5, 1, 0)
confusionMatrix(test_predictions, test_data$d)
Reference
Prediction 0 1
0 187 0
1 0 60