confusionMatrix Error in newdata.df$Propensity : $ operator is invalid for atomic vectors

Question

A resulting set of predictive model validation for a classification model includes both actual values and propensities.

Create a data frame using cbind() command, include Propensity and Actual values in the data frame.

Propensity <- c(0.53, 0.42, 0.38, 0.76, 0.33, 0.42, 0.55, 0.59, 0.09, 0.21, 0.32, 0.52, 0.03, 0.13, 0.01, 0.68, 0.48, 0.29, 0.03, 0.02)
Actual <- c(1, 1, 0, 1, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0)
newdata.df <- cbind(Propensity,Actual)
confusionMatrix(as.factor(ifelse(newdata.df$Propensity > 0.5, 'Propensity', 'Actual')),newdata.df$Actual)

Answer 1

First, you created a matrix with cbind() but the $colname works only with data frames. Second, confusionMatrix wants two factors that have the same levels, both 1, 0 or both Actual, Propensity:

library(caret)  # Always include packages loaded. The `confusionMatrix` function is not part of base R
Prop <- as.factor(ifelse(newdata.df$Propensity > 0.5, 'Propensity', 'Actual'))
Act <- as.factor(ifelse(newdata.df$Actual > 0.5, 'Propensity', 'Actual'))
newdata.df <- data.frame(Prop, Act)
confusionMatrix(newdata.df$Prop, newdata.df$Act)
# Confusion Matrix and Statistics
# 
#             Reference
# Prediction   Actual Propensity
#   Actual         10          4
#   Propensity      1          5
#                                          
#                Accuracy : 0.75           
#                  95% CI : (0.509, 0.9134)
#     No Information Rate : 0.55           
#     P-Value [Acc > NIR] : 0.05533        
#                                          
#                   Kappa : 0.4792         
#                                          
#  Mcnemar's Test P-Value : 0.37109        
#                                          
#             Sensitivity : 0.9091         
#             Specificity : 0.5556         
#          Pos Pred Value : 0.7143         
#          Neg Pred Value : 0.8333         
#              Prevalence : 0.5500         
#          Detection Rate : 0.5000         
#    Detection Prevalence : 0.7000         
#       Balanced Accuracy : 0.7323         
#                                          
#        'Positive' Class : Actual