Support some parameters for ALS.train() in Python

Author: yu-iskw

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

@@ -506,11 +506,21 @@ private[python] class PythonMLLibAPI extends Serializable {
    * Java stub for Python mllib LDA.run()
    */
   def trainLDAModel(
-    data: JavaRDD[LabeledPoint],
-    k: Int,
-    seed: java.lang.Long): LDAModel = {
+      data: JavaRDD[LabeledPoint],
+      k: Int,
+      maxIterations: Int,
+      docConcentration: Double,
+      topicConcentration: Double,
+      seed: java.lang.Long,
+      checkpointInterval: Int,
+      optimizer: String): LDAModel = {
     val algo = new LDA()
         .setK(k)
+        .setMaxIterations(maxIterations)
+        .setDocConcentration(docConcentration)
+        .setTopicConcentration(topicConcentration)
+        .setCheckpointInterval(checkpointInterval)
+        .setOptimizer(optimizer)
 
     if (seed != null) algo.setSeed(seed)
 

python/pyspark/mllib/clustering.py

@@ -597,7 +597,7 @@ class LDAModel(JavaModelWrapper):
     ...     LabeledPoint(2, [1.0, 0.0]),
     ... ]
     >>> rdd =  sc.parallelize(data)
-    >>> model = LDA.train(rdd, 2)
+    >>> model = LDA.train(rdd, k=2)
     >>> model.vocabSize()
     2
     >>> topics = model.topicsMatrix()
@@ -625,8 +625,12 @@ def describeTopics(self, maxTermsPerTopic=None):
 class LDA():
 
     @classmethod
-    def train(cls, rdd, k, seed=None):
-        model = callMLlibFunc("trainLDAModel", rdd, k, seed)
+    def train(cls, rdd, k=10, maxIterations=20, docConcentration=-1.0,
+              topicConcentration=-1.0, seed=None, checkpointInterval=10, optimizer="em"):
+        """Train a LDA model."""
+        model = callMLlibFunc("trainLDAModel", rdd, k, maxIterations,
+                              docConcentration, topicConcentration, seed,
+                              checkpointInterval, optimizer)
         return LDAModel(model)