In-sample fit full model
It is coding time again, which means coming back to the exercise dataset defaultData
.
You now want to know how your model performs by calculating the accuracy. In order to do so, you first need a confusion matrix.
Take the logitModelFull
, first. The model is already specified and lives in your environment.
This exercise is part of the course
Machine Learning for Marketing Analytics in R
Exercise instructions
- Use
predict()
to receive a probability of each customer defaulting on their payment. - In order to construct the confusion matrix use the function
confusion.matrix()
fromSDMTools
. Note thatSDMTools
cannot be downloaded from CRAN anymore. So if you want to practice at your home computer you can install the package by usingremotes::install_version("SDMTools", "1.1-221.2")
which will install the version ofSDMTools
that is used at this course. - Choose a common threshold of 0.5.
- Calculate the accuracy using the confusion matrix.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Make predictions using the full Model
defaultData$predFull <- predict(logitModelFull, type = ___, na.action = ___)
# Construct the in-sample confusion matrix
confMatrixModelFull <- confusion.matrix(defaultData$___,defaultData$___, threshold = ___)
confMatrixModelFull
# Calculate the accuracy for the full Model
accuracyFull <- sum(diag(___)) / ___(___)
accuracyFull