As mentioned before, model evaluation is built-in to ApacheSparkML and you'll find all that you need in the org.apache.spark.ml.evaluation package. Let's continue with our binary classification. This means that we'll have to use org.apache.spark.ml.evaluation.BinaryClassificationEvaluator:

import org.apache.spark.ml.evaluation.BinaryClassificationEvaluator
val evaluator = new BinaryClassificationEvaluator()

import org.apache.spark.ml.param.ParamMap
var evaluatorParamMap = ParamMap(evaluator.metricName -> "areaUnderROC")
var aucTraining = evaluator.evaluate(result, evaluatorParamMap)

To code previous initialized a BinaryClassificationEvaluator function and tells it to calculate the areaUnderROC, one of the many possible metrics to assess the prediction performance of a machine learning algorithm.

As we have the actual label and the prediction present in a DataFrame called result, it is simple to calculate this score and is done using the following line of code:

var aucTraining = evaluator.evaluate(result, evaluatorParamMap)