Styjun

I am currently clustering some text documents.
I am using K-means and proceed my data with TF-IDF thanks to the PySpark methods.
And now I want to get the top 10 words for each cluster :

When I do :

getTopwords_udf = udf(lambda vector: [ countVectorizerModel.vocabulary[indice] for indice in vector.toArray().tolist().argsort()[-10:][::-1]], ArrayType(StringType()))

predictions.groupBy("prediction").agg(Summarizer.mean(col("features")).alias("means")) 
 .withColumn("topWord", getTopwords_udf(col('means'))) 
 .select("prediction", "topWord") 
 .show(2, truncate=100)

I am getting this error :

Could not serialize object: Py4JError: An error occurred while calling o225.__getstate__. Trace:
py4j.Py4JException: Method __getstate__([]) does not exist
 at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:318)
 at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:326)
 at py4j.Gateway.invoke(Gateway.java:274)
 at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)
 at py4j.commands.CallCommand.execute(CallCommand.java:79)
 at py4j.GatewayConnection.run(GatewayConnection.java:238)
 at java.lang.Thread.run(Thread.java:748)


Traceback (most recent call last):
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/sql/udf.py", line 189, in wrapper
 return self(*args)
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/sql/udf.py", line 167, in __call__
 judf = self._judf
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/sql/udf.py", line 151, in _judf
 self._judf_placeholder = self._create_judf()
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/sql/udf.py", line 160, in _create_judf
 wrapped_func = _wrap_function(sc, self.func, self.returnType)
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/sql/udf.py", line 35, in _wrap_function
 pickled_command, broadcast_vars, env, includes = _prepare_for_python_RDD(sc, command)
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/rdd.py", line 2420, in _prepare_for_python_RDD
 pickled_command = ser.dumps(command)
 File "/opt/bigpipe/spark/python/lib/pyspark.zip/pyspark/serializers.py", line 597, in dumps
 raise pickle.PicklingError(msg)
_pickle.PicklingError: Could not serialize object: Py4JError: An error occurred while calling o225.__getstate__. Trace:
py4j.Py4JException: Method __getstate__([]) does not exist
 at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:318)
 at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:326)
 at py4j.Gateway.invoke(Gateway.java:274)
 at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)
 at py4j.commands.CallCommand.execute(CallCommand.java:79)
 at py4j.GatewayConnection.run(GatewayConnection.java:238)
 at java.lang.Thread.run(Thread.java:748)

I thought it was because of the type (from DoubleType to float for numpy) so I have tried this as well to see what is happening

vector_udf = udf(lambda vector: vector.toArray().tolist(), ArrayType(FloatType()))
vector2_udf = udf(lambda vector: vector.sort()[:10], ArrayType(FloatType()))

predictions.groupBy("prediction").agg(Summarizer.mean(col("features")).alias("means")) 
 .withColumn("topWord", vector_udf(col('means'))) 
 .withColumn("topWord2", vector2_udf(col('topWord'))) 
 .select("prediction", "topWord", "topWord2") 
 .show(2, truncate=100)

But I get this error TypeError: 'NoneType' object is not subscriptable

I have figured out how to get the top X of words from a SparseVector to a string array with PySpark.
Here is my solution for those who might be interested...

def getTopWordContainer(v):
 def getTopWord(vector):
 vectorConverted = vector.toArray().tolist()
 listSortedDesc= [i[0] for i in sorted(enumerate(vectorConverted), key=lambda x:x[1])][-10:][::-1]
 return [v[j] for j in listSortedDesc]
 return getTopWord

getTopWordInit = getTopWordContainer(countVectorizerModel.vocabulary)
getTopWord_udf = udf(getTopWordInit, ArrayType(StringType()))

top = predictions.groupBy("prediction").agg(Summarizer.mean(col("features")).alias("means")) 
 .withColumn("topWord", getTopWord_udf(col('means'))) 
 .select("prediction", "topWord")

I am a beginner in spark so if you know hot to enhance it, let me know :)