predict_new() Forces Target Variable To Be Present In The New Data

[drag05]

Running the example under predict_new() documentation with the "Price" target variable removed from new data throws this error:

> out    <- train(lisbon_train, 'Price', verbose = FALSE, bayes_iter = 0, random_evals = 1)

# here, lisbon_new does not have the "Price":

> preds  <- predict_new(out, lisbon_new)

Error in `[.data.frame`(xgboost_data, , y) : undefined columns selected

> traceback()

5: stop("undefined columns selected")
4: `[.data.frame`(xgboost_data, , y)
3: xgboost_data[, y]
2: prepare_data(data, type = type, y = y, engine = engine, predict = TRUE, 
       train = train_data)
1: predict_new(out, lisbon_new)

prepare_data() seems to be y-aware and does not work without the target variable. However, having the target values present in the new data renders prediction unnecessary.

Thank you!