Remove unnecessary appendBias implementation

Author: Lewuathe

mllib/src/main/scala/org/apache/spark/mllib/api/python/PythonMLLibAPI.scala

@@ -71,13 +71,15 @@ private[python] class PythonMLLibAPI extends Serializable {
       minPartitions: Int): JavaRDD[LabeledPoint] =
     MLUtils.loadLabeledPoints(jsc.sc, path, minPartitions)
 
-  def appendBias(data: org.apache.spark.mllib.linalg.Vector)
-  : org.apache.spark.mllib.linalg.Vector
-    = MLUtils.appendBias(data)
-
-  def loadVectors(jsc: JavaSparkContext, path: String)
-  : RDD[org.apache.spark.mllib.linalg.Vector]
-    = MLUtils.loadVectors(jsc.sc, path)
+  /**
+   * Loads and serializes vectors saved with `RDD#saveAsTextFile`.
+   * @param jsc Java SparkContext
+   * @param path file or directory path in any Hadoop-supported file system URI
+   * @return serialized vectors in a RDD
+   */
+  def loadVectors(jsc: JavaSparkContext,
+                  path: String): RDD[Vector] =
+    MLUtils.loadVectors(jsc.sc, path)
 
   private def trainRegressionModel(
       learner: GeneralizedLinearAlgorithm[_ <: GeneralizedLinearModel],

python/pyspark/mllib/tests.py

@@ -821,7 +821,7 @@ def test_model_transform(self):
 
 class MLUtilsTests(MLlibTestCase):
     def test_append_bias(self):
-        data = [1.0, 2.0, 3.0]
+        data = [2.0, 2.0, 2.0]
         ret = MLUtils.appendBias(data)
         self.assertEqual(ret[3], 1.0)
 
@@ -832,14 +832,17 @@ def test_load_vectors(self):
             [1.0, 2.0, 3.0]
         ]
         try:
-            self.sc.parallelize(data).saveAsTextFile("test_load_vectors")
-            ret_rdd = MLUtils.loadVectors(self.sc, "test_load_vectors")
+            temp_dir = tempfile.mkdtemp()
+            load_vectors_path = os.path.join(temp_dir, "test_load_vectors")
+            self.sc.parallelize(data).saveAsTextFile(load_vectors_path)
+            ret_rdd = MLUtils.loadVectors(self.sc, load_vectors_path)
             ret = ret_rdd.collect()
+            ret.sort()
             self.assertEqual(len(ret), 2)
             self.assertEqual(ret[0], DenseVector([1.0, 2.0, 3.0]))
             self.assertEqual(ret[1], DenseVector([1.0, 2.0, 3.0]))
         finally:
-            shutil.rmtree("test_load_vectors")
+            shutil.rmtree(load_vectors_path)
 
 
 if __name__ == "__main__":

python/pyspark/mllib/util.py

@@ -171,10 +171,18 @@ def loadLabeledPoints(sc, path, minPartitions=None):
 
     @staticmethod
     def appendBias(data):
-        return callMLlibFunc("appendBias", _convert_to_vector(data))
+        """
+        Returns a new vector with `1.0` (bias) appended to the input vector.
+        """
+        vec = _convert_to_vector(data)
+        return np.append(vec, 1.0)
 
     @staticmethod
     def loadVectors(sc, path):
+        """
+        Loads vectors saved using `RDD[Vector].saveAsTextFile`
+        with the default number of partitions.
+        """
         return callMLlibFunc("loadVectors", sc, path)