Solved – Understanding factors returned by RPart classification

rrpart

This is a very basic question on using R for classification. I'm trying to use rpart for classification task and would like to have a class label as a result, i.e. I use type="class" in predict method

predict(tree, data, type = "class")

The class labels are "0" and "1". When I try to classify a single observation I get something like this:

1 
0 
Levels: 0 1

As I understand this is a factor with 2 levels: 0 and 1. However, why there are 2 numbers reported and what is exactly the class assigned to this observation?

Best Answer

You didn't show a dataset to replicate. You can see from a basic illustration that class typically only returns one class per prediction. Maybe in your case, 1 is the observation index and 0 is the prediction with 2 corresponding factor levels possible {0,1}.

   > predict(fit, type = "class")  # factor
          1       2       3       4       5       6       7       8       9      10 
    present  absent present present  absent  absent  absent  absent  absent present 
         11      12      13      14      15      16      17      18      19      20 
    present  absent present  absent  absent  absent  absent  absent  absent  absent 
     21      22      23      24      25      26      27      28      29      30 

Levels: absent present

You can also store the result in a variable and look at the structure of that variable to confirm the specific output objects using str() function. Notice the prediction attribute values and Factor names attributes are clearly specified here.

 p<-predict(fit, type = "class")
 str(p)
 Factor w/ 2 levels "absent","present": 2 1 2 2 1 1 1 1 1 2 ...
 - attr(*, "names")= chr [1:81] "1" "2" "3" "4" ...

Related Solutions

Rpart Prediction – How to Use Rpart’s Result for Classification

fit = rpart(formula, data =, method =, control=)
fitVariablesUsed <- names(fit[,1:20])
preds <- predict(fit, data = newdata[,c(fitVariablesUsed)], type = c("prob"))

this will return a probability matrix for each of the observations. Meaning it will give a probability that the observation is in class 1, class2, etc.

make sure that the columns all line up correctly between the matrix you made the model with and the matrix you're going to make the predictions with.

The variables I created fitVaraiblesUsed which connected 20 variables (just for example) from the fit data frame can then be used in the new data data frame, so long as they're all named the same thing.

Solved – rpart classification: why is the predict() output not adhering to type=”class”

Predicting with an rpart object also includes the parameter Type. From the help:

If the rpart object is a classification tree, then the default is to return prob predictions, a matrix whose columns are the probability of the first, second, etc. class.

In your case it returns the probabilities of the classes, not the class itself. Use the following code to get the classes.

pred = predict(fit, type = "class")
table(pred, datas$bad_econ)

for more information on residual mean deviance check this post on cv

Edit: For measuring AUC and plot the ROC, you could use these commands:

library(ROCR)
rocrpred <- prediction(pred[,2], data$bad_econ)
# print AUC value
as.numeric(performance(rocrpred, "auc")@y.values)
# plots the ROC curve with colors where the splits are.
plot(performance(rocrpred, "tpr", "fpr"), colorize = TRUE)

But read the help for the performance function. Because it can also create lift charts, Sensitivity/specificity plots, or Precision/recall graphs.

I very seldom look at the residual mean deviance. But I compare the confusion matrix of the prediction model with the confusion matrix of a baseline model. Same goes for the roc curve and the auc values. Those are easier to explain to non-stats people.

Best Answer

Related Solutions

Rpart Prediction – How to Use Rpart’s Result for Classification

Solved – rpart classification: why is the predict() output not adhering to type=”class”

Related Question