Working with the classifyr Package

Use the example package, even without installing

We use export_all = FALSE to simulate how R behaves when calling library.

devtools::load_all("classifyr", export_all = FALSE)

ℹ Loading classifyr

Train the model

We have a simple function to “train” a two-stage model:

model <- train_classifyr(iris, target = "Species")

Training classifyr model

Bundling models into a single object

This object is a list with a custom class attribute tacked onto the front:

attr(model, which = "class")

[1] "classifyr_model" "list"           

Because we’ve defined a custom print method, though—print.classifyr_model()—we can control how the object is printed or displayed in settings where a string representation is needed:

model

Classifyr model
- Target: Species 
- Features: Sepal.Length, Sepal.Width, Petal.Length, Petal.Width 

Generate predictions

We have defined a predict method as well, so we can use the model just like any other in R:

preds <- predict(model, iris)

Making predictions for classifyr model

head(preds, n = 3)

[1] 0.2071281 0.6798370 0.7549796

Explain the model

Finally, maybe we’d like to add other functionality. Here we’ve (1) created and (2) defined a new explain method:

explns <- explain(model, preds, iris)
head(explns, n = 5)

Explaining model predictions

       pred letter number                                          message
1 0.2071281      K     22 Brought to you by the letter K and the number 22
2 0.6798370      V     71 Brought to you by the letter V and the number 71
3 0.7549796      Q     52 Brought to you by the letter Q and the number 52
4 0.5999480      X     31 Brought to you by the letter X and the number 31
5 0.4022563      U     13 Brought to you by the letter U and the number 13