[SPARK-14564][ML][MLLIB][PYSPARK] Python Word2Vec missing setWindowSi…

…ze method

## What changes were proposed in this pull request?
Added windowSize getter/setter to ML/MLlib

## How was this patch tested?
Added test cases in tests.py under both ML and MLlib

Author: Jason Lee <cjlee@us.ibm.com>

Closes #12428 from jasoncl/SPARK-14564.

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

@@ -671,6 +671,7 @@ private[python] class PythonMLLibAPI extends Serializable {
    * @param numPartitions number of partitions
    * @param numIterations number of iterations
    * @param seed initial seed for random generator
+   * @param windowSize size of window
    * @return A handle to java Word2VecModelWrapper instance at python side
    */
   def trainWord2VecModel(
@@ -680,14 +681,16 @@ private[python] class PythonMLLibAPI extends Serializable {
       numPartitions: Int,
       numIterations: Int,
       seed: Long,
-      minCount: Int): Word2VecModelWrapper = {
+      minCount: Int,
+      windowSize: Int): Word2VecModelWrapper = {
     val word2vec = new Word2Vec()
       .setVectorSize(vectorSize)
       .setLearningRate(learningRate)
       .setNumPartitions(numPartitions)
       .setNumIterations(numIterations)
       .setSeed(seed)
       .setMinCount(minCount)
+      .setWindowSize(windowSize)
     try {
       val model = word2vec.fit(dataJRDD.rdd.persist(StorageLevel.MEMORY_AND_DISK_SER))
       new Word2VecModelWrapper(model)

python/pyspark/ml/feature.py

@@ -2219,28 +2219,31 @@ class Word2Vec(JavaEstimator, HasStepSize, HasMaxIter, HasSeed, HasInputCol, Has
     minCount = Param(Params._dummy(), "minCount",
                      "the minimum number of times a token must appear to be included in the " +
                      "word2vec model's vocabulary", typeConverter=TypeConverters.toInt)
+    windowSize = Param(Params._dummy(), "windowSize",
+                       "the window size (context words from [-window, window]). Default value is 5",
+                       typeConverter=TypeConverters.toInt)
 
     @keyword_only
     def __init__(self, vectorSize=100, minCount=5, numPartitions=1, stepSize=0.025, maxIter=1,
-                 seed=None, inputCol=None, outputCol=None):
+                 seed=None, inputCol=None, outputCol=None, windowSize=5):
         """
         __init__(self, vectorSize=100, minCount=5, numPartitions=1, stepSize=0.025, maxIter=1, \
-                 seed=None, inputCol=None, outputCol=None)
+                 seed=None, inputCol=None, outputCol=None, windowSize=5)
         """
         super(Word2Vec, self).__init__()
         self._java_obj = self._new_java_obj("org.apache.spark.ml.feature.Word2Vec", self.uid)
         self._setDefault(vectorSize=100, minCount=5, numPartitions=1, stepSize=0.025, maxIter=1,
-                         seed=None)
+                         seed=None, windowSize=5)
         kwargs = self.__init__._input_kwargs
         self.setParams(**kwargs)
 
     @keyword_only
     @since("1.4.0")
     def setParams(self, vectorSize=100, minCount=5, numPartitions=1, stepSize=0.025, maxIter=1,
-                  seed=None, inputCol=None, outputCol=None):
+                  seed=None, inputCol=None, outputCol=None, windowSize=5):
         """
         setParams(self, minCount=5, numPartitions=1, stepSize=0.025, maxIter=1, seed=None, \
-                 inputCol=None, outputCol=None)
+                 inputCol=None, outputCol=None, windowSize=5)
         Sets params for this Word2Vec.
         """
         kwargs = self.setParams._input_kwargs
@@ -2291,6 +2294,21 @@ def getMinCount(self):
         """
         return self.getOrDefault(self.minCount)
 
+    @since("2.0.0")
+    def setWindowSize(self, value):
+        """
+        Sets the value of :py:attr:`windowSize`.
+        """
+        self._set(windowSize=value)
+        return self
+
+    @since("2.0.0")
+    def getWindowSize(self):
+        """
+        Gets the value of windowSize or its default value.
+        """
+        return self.getOrDefault(self.windowSize)
+
     def _create_model(self, java_model):
         return Word2VecModel(java_model)
 

python/pyspark/ml/tests.py

@@ -341,6 +341,11 @@ def test_param_property_error(self):
         params = param_store.params  # should not invoke the property 'test_property'
         self.assertEqual(len(params), 1)
 
+    def test_word2vec_param(self):
+        model = Word2Vec().setWindowSize(6)
+        # Check windowSize is set properly
+        self.assertEqual(model.getWindowSize(), 6)
+
 
 class FeatureTests(PySparkTestCase):
 

python/pyspark/mllib/feature.py

@@ -617,6 +617,7 @@ def __init__(self):
         self.numIterations = 1
         self.seed = random.randint(0, sys.maxsize)
         self.minCount = 5
+        self.windowSize = 5
 
     @since('1.2.0')
     def setVectorSize(self, vectorSize):
@@ -669,6 +670,14 @@ def setMinCount(self, minCount):
         self.minCount = minCount
         return self
 
+    @since('2.0.0')
+    def setWindowSize(self, windowSize):
+        """
+        Sets window size (default: 5).
+        """
+        self.windowSize = windowSize
+        return self
+
     @since('1.2.0')
     def fit(self, data):
         """
@@ -682,7 +691,7 @@ def fit(self, data):
         jmodel = callMLlibFunc("trainWord2VecModel", data, int(self.vectorSize),
                                float(self.learningRate), int(self.numPartitions),
                                int(self.numIterations), int(self.seed),
-                               int(self.minCount))
+                               int(self.minCount), int(self.windowSize))
         return Word2VecModel(jmodel)
 
 

python/pyspark/mllib/tests.py

@@ -1027,13 +1027,15 @@ def test_word2vec_setters(self):
             .setNumPartitions(2) \
             .setNumIterations(10) \
             .setSeed(1024) \
-            .setMinCount(3)
+            .setMinCount(3) \
+            .setWindowSize(6)
         self.assertEqual(model.vectorSize, 2)
         self.assertTrue(model.learningRate < 0.02)
         self.assertEqual(model.numPartitions, 2)
         self.assertEqual(model.numIterations, 10)
         self.assertEqual(model.seed, 1024)
         self.assertEqual(model.minCount, 3)
+        self.assertEqual(model.windowSize, 6)
 
     def test_word2vec_get_vectors(self):
         data = [