See the Spark ML Documentation MulticlassClassificationEvaluator
ml_classification_eval(predicted_tbl_spark, label, predicted_lbl,
metric = "f1")
A tbl_spark object that contains a columns with predicted labels
Name of the column that contains the true, indexed label. Support for binary and multi-class labels, column should be of double type (use as.double)
Name of the column that contains the predicted label NOT the scored probability. Support for binary and multi-class labels, column should be of double type (use as.double)
A classification metric, for Spark 1.6: f1 (default), precision, recall, weightedPrecision, weightedRecall or accuracy; for Spark 2.X: f1 (default), weightedPrecision, weightedRecall or accuracy
see metric