pyspark : NameError: name 'spark' is not defined

I am copying the pyspark.ml example from the official document website:

data = [(Vectors.dense([0.0, 0.0]),), (Vectors.dense([1.0, 1.0]),),(Vectors.dense([9.0, 8.0]),), (Vectors.dense([8.0, 9.0]),)] df = spark.createDataFrame(data, ["features"]) kmeans = KMeans(k=2, seed=1) model = kmeans.fit(df) 

However, the example above wouldn't run and gave me the following errors:

--------------------------------------------------------------------------- NameError Traceback (most recent call last) <ipython-input-28-aaffcd1239c9> in <module>() 1 from pyspark import * 2 data = [(Vectors.dense([0.0, 0.0]),), (Vectors.dense([1.0, 1.0]),),(Vectors.dense([9.0, 8.0]),), (Vectors.dense([8.0, 9.0]),)] ----> 3 df = spark.createDataFrame(data, ["features"]) 4 kmeans = KMeans(k=2, seed=1) 5 model = kmeans.fit(df) NameError: name 'spark' is not defined 

What additional configuration/variable needs to be set to get the example running?

5 Answers

You can add

from pyspark.context import SparkContext from pyspark.sql.session import SparkSession sc = SparkContext('local') spark = SparkSession(sc) 

to the begining of your codes to define a SparkSession, then the spark.createDataFrame() should work.

Answer by 率怀一 is good and will work for the first time. But the second time you try it, it will throw the following exception :

ValueError: Cannot run multiple SparkContexts at once; existing SparkContext(app=pyspark-shell, master=local) created by __init__ at <ipython-input-3-786525f7559f>:10 

There are two ways to avoid it.

1) Using SparkContext.getOrCreate() instead of SparkContext():

from pyspark.context import SparkContext from pyspark.sql.session import SparkSession sc = SparkContext.getOrCreate() spark = SparkSession(sc) 

2) Using sc.stop() in the end, or before you start another SparkContext.

Since you are calling createDataFrame(), you need to do this:

df = sqlContext.createDataFrame(data, ["features"]) 

instead of this:

df = spark.createDataFrame(data, ["features"]) 

spark stands there as the sqlContext.

In general, some people have that as sc, so if that didn't work, you could try:

df = sc.createDataFrame(data, ["features"]) 

You have to import the spark as following if you are using python then it will create a spark session but remember it is an old method though it will work.

from pyspark.shell import spark 

If it errors you regarding other open session do this:

from pyspark.context import SparkContext from pyspark.sql.session import SparkSession sc = SparkContext.getOrCreate(); spark = SparkSession(sc) scraped_data=spark.read.json("/Users/reihaneh/Desktop/nov3_final_tst1/")