diff --git a/mllib/src/main/scala/org/apache/spark/mllib/regression/Lasso.scala b/mllib/src/main/scala/org/apache/spark/mllib/regression/Lasso.scala
index 911e0c0178cab..25920d0dc976e 100644
--- a/mllib/src/main/scala/org/apache/spark/mllib/regression/Lasso.scala
+++ b/mllib/src/main/scala/org/apache/spark/mllib/regression/Lasso.scala
@@ -17,13 +17,11 @@
 
 package org.apache.spark.mllib.regression
 
-import breeze.linalg.{Vector => BV}
-
 import org.apache.spark.SparkContext
-import org.apache.spark.rdd.RDD
-import org.apache.spark.mllib.linalg.{Vector, Vectors}
+import org.apache.spark.mllib.linalg.Vector
 import org.apache.spark.mllib.optimization._
 import org.apache.spark.mllib.util.MLUtils
+import org.apache.spark.rdd.RDD
 
 /**
  * Regression model trained using Lasso.
@@ -58,8 +56,7 @@ class LassoWithSGD private (
     var numIterations: Int,
     var regParam: Double,
     var miniBatchFraction: Double)
-  extends GeneralizedLinearAlgorithm[LassoModel]
-  with Serializable {
+  extends GeneralizedLinearAlgorithm[LassoModel] with Serializable {
 
   val gradient = new LeastSquaresGradient()
   val updater = new L1Updater()
@@ -71,10 +68,6 @@ class LassoWithSGD private (
   // We don't want to penalize the intercept, so set this to false.
   super.setIntercept(false)
 
-  private var yMean = 0.0
-  private var xColMean: BV[Double] = _
-  private var xColSd: BV[Double] = _
-
   /**
    * Construct a Lasso object with default parameters
    */
@@ -87,31 +80,7 @@ class LassoWithSGD private (
   }
 
   override protected def createModel(weights: Vector, intercept: Double) = {
-    val weightsMat = weights.toBreeze
-    val weightsScaled = weightsMat :/ xColSd
-    val interceptScaled = yMean - weightsMat.dot(xColMean :/ xColSd)
-
-    new LassoModel(Vectors.fromBreeze(weightsScaled), interceptScaled)
-  }
-
-  override def run(input: RDD[LabeledPoint], initialWeights: Vector): LassoModel = {
-    val nfeatures: Int = input.first.features.size
-    val nexamples: Long = input.count()
-
-    // To avoid penalizing the intercept, we center and scale the data.
-    val stats = MLUtils.computeStats(input, nfeatures, nexamples)
-    yMean = stats._1
-    xColMean = stats._2.toBreeze
-    xColSd = stats._3.toBreeze
-
-    val normalizedData = input.map { point =>
-      val yNormalized = point.label - yMean
-      val featuresMat = point.features.toBreeze
-      val featuresNormalized = (featuresMat - xColMean) :/ xColSd
-      LabeledPoint(yNormalized, Vectors.fromBreeze(featuresNormalized))
-    }
-
-    super.run(normalizedData, initialWeights)
+    new LassoModel(weights, intercept)
   }
 }
 
diff --git a/mllib/src/test/java/org/apache/spark/mllib/regression/JavaLassoSuite.java b/mllib/src/test/java/org/apache/spark/mllib/regression/JavaLassoSuite.java
index f44b25cd44d19..f725924a2d971 100644
--- a/mllib/src/test/java/org/apache/spark/mllib/regression/JavaLassoSuite.java
+++ b/mllib/src/test/java/org/apache/spark/mllib/regression/JavaLassoSuite.java
@@ -59,7 +59,7 @@ int validatePrediction(List<LabeledPoint> validationData, LassoModel model) {
   @Test
   public void runLassoUsingConstructor() {
     int nPoints = 10000;
-    double A = 2.0;
+    double A = 0.0;
     double[] weights = {-1.5, 1.0e-2};
 
     JavaRDD<LabeledPoint> testRDD = sc.parallelize(LinearDataGenerator.generateLinearInputAsList(A,
@@ -80,7 +80,7 @@ public void runLassoUsingConstructor() {
   @Test
   public void runLassoUsingStaticMethods() {
     int nPoints = 10000;
-    double A = 2.0;
+    double A = 0.0;
     double[] weights = {-1.5, 1.0e-2};
 
     JavaRDD<LabeledPoint> testRDD = sc.parallelize(LinearDataGenerator.generateLinearInputAsList(A,
diff --git a/mllib/src/test/scala/org/apache/spark/mllib/regression/LassoSuite.scala b/mllib/src/test/scala/org/apache/spark/mllib/regression/LassoSuite.scala
index 02e99fac6146f..bfa42959c8ead 100644
--- a/mllib/src/test/scala/org/apache/spark/mllib/regression/LassoSuite.scala
+++ b/mllib/src/test/scala/org/apache/spark/mllib/regression/LassoSuite.scala
@@ -40,10 +40,11 @@ class LassoSuite extends FunSuite with LocalSparkContext {
     val B = -1.5
     val C = 1.0e-2
 
-    val testData = LinearDataGenerator.generateLinearInput(A, Array[Double](B,C), nPoints, 42)
-
-    val testRDD = sc.parallelize(testData, 2)
-    testRDD.cache()
+    val testData = LinearDataGenerator.generateLinearInput(A, Array[Double](B, C), nPoints, 42)
+      .map { case LabeledPoint(label, features) =>
+      LabeledPoint(label, Vectors.dense(1.0 +: features.toArray))
+    }
+    val testRDD = sc.parallelize(testData, 2).cache()
 
     val ls = new LassoWithSGD()
     ls.optimizer.setStepSize(1.0).setRegParam(0.01).setNumIterations(40)
@@ -51,11 +52,15 @@ class LassoSuite extends FunSuite with LocalSparkContext {
     val model = ls.run(testRDD)
     val weight0 = model.weights(0)
     val weight1 = model.weights(1)
-    assert(model.intercept >= 1.9 && model.intercept <= 2.1, model.intercept + " not in [1.9, 2.1]")
-    assert(weight0 >= -1.60 && weight0 <= -1.40, weight0 + " not in [-1.6, -1.4]")
-    assert(weight1 >= -1.0e-3 && weight1 <= 1.0e-3, weight1 + " not in [-0.001, 0.001]")
+    val weight2 = model.weights(2)
+    assert(weight0 >= 1.9 && weight0 <= 2.1, weight0 + " not in [1.9, 2.1]")
+    assert(weight1 >= -1.60 && weight1 <= -1.40, weight1 + " not in [-1.6, -1.4]")
+    assert(weight2 >= -1.0e-3 && weight2 <= 1.0e-3, weight2 + " not in [-0.001, 0.001]")
 
     val validationData = LinearDataGenerator.generateLinearInput(A, Array[Double](B,C), nPoints, 17)
+      .map { case LabeledPoint(label, features) =>
+      LabeledPoint(label, Vectors.dense(1.0 +: features.toArray))
+    }
     val validationRDD  = sc.parallelize(validationData, 2)
 
     // Test prediction on RDD.
@@ -73,13 +78,16 @@ class LassoSuite extends FunSuite with LocalSparkContext {
     val C = 1.0e-2
 
     val testData = LinearDataGenerator.generateLinearInput(A, Array[Double](B, C), nPoints, 42)
+      .map { case LabeledPoint(label, features) =>
+      LabeledPoint(label, Vectors.dense(1.0 +: features.toArray))
+    }
 
+    val initialA = -1.0
     val initialB = -1.0
     val initialC = -1.0
-    val initialWeights = Vectors.dense(initialB, initialC)
+    val initialWeights = Vectors.dense(initialA, initialB, initialC)
 
-    val testRDD = sc.parallelize(testData, 2)
-    testRDD.cache()
+    val testRDD = sc.parallelize(testData, 2).cache()
 
     val ls = new LassoWithSGD()
     ls.optimizer.setStepSize(1.0).setRegParam(0.01).setNumIterations(40)
@@ -87,11 +95,15 @@ class LassoSuite extends FunSuite with LocalSparkContext {
     val model = ls.run(testRDD, initialWeights)
     val weight0 = model.weights(0)
     val weight1 = model.weights(1)
-    assert(model.intercept >= 1.9 && model.intercept <= 2.1, model.intercept + " not in [1.9, 2.1]")
-    assert(weight0 >= -1.60 && weight0 <= -1.40, weight0 + " not in [-1.6, -1.4]")
-    assert(weight1 >= -1.0e-3 && weight1 <= 1.0e-3, weight1 + " not in [-0.001, 0.001]")
+    val weight2 = model.weights(2)
+    assert(weight0 >= 1.9 && weight0 <= 2.1, weight0 + " not in [1.9, 2.1]")
+    assert(weight1 >= -1.60 && weight1 <= -1.40, weight1 + " not in [-1.6, -1.4]")
+    assert(weight2 >= -1.0e-3 && weight2 <= 1.0e-3, weight2 + " not in [-0.001, 0.001]")
 
     val validationData = LinearDataGenerator.generateLinearInput(A, Array[Double](B,C), nPoints, 17)
+      .map { case LabeledPoint(label, features) =>
+      LabeledPoint(label, Vectors.dense(1.0 +: features.toArray))
+    }
     val validationRDD  = sc.parallelize(validationData,2)
 
     // Test prediction on RDD.