Return Array[Double] from SGD instead of DoubleMatrix

mllib/src/main/scala/spark/mllib/optimization/GradientDescent.scala

@@ -50,7 +50,7 @@ object GradientDescent {
     stepSize: Double,
     numIters: Int,
     initialWeights: Array[Double],
-    miniBatchFraction: Double=1.0) : (DoubleMatrix, Array[Double]) = {
+    miniBatchFraction: Double=1.0) : (Array[Double], Array[Double]) = {
 
     val stochasticLossHistory = new ArrayBuffer[Double](numIters)
 
@@ -75,6 +75,6 @@ object GradientDescent {
       reg_val = update._2
     }
 
-    (weights, stochasticLossHistory.toArray)
+    (weights.toArray, stochasticLossHistory.toArray)
   }
 }

mllib/src/main/scala/spark/mllib/regression/LogisticRegression.scala

@@ -126,10 +126,8 @@ class LogisticRegression private (var stepSize: Double, var miniBatchFraction: D
       initalWeightsWithIntercept,
       miniBatchFraction)
 
-    val weightsArray = weights.toArray()
-
-    val intercept = weightsArray(0)
-    val weightsScaled = weightsArray.tail
+    val intercept = weights(0)
+    val weightsScaled = weights.tail
 
     val model = new LogisticRegressionModel(weightsScaled, intercept, stochasticLosses)