Apache spark dealing with case statements

Kunal Anand picture Kunal Anand · Oct 11, 2016 · Viewed 60.3k times · Source

I am dealing with transforming SQL code to PySpark code and came across some SQL statements. I don't know how to approach case statments in pyspark? I am planning on creating a RDD and then using rdd.map and then do some logic checks. Is that the right approach? Please help!

Basically I need to go through each line in the RDD or DF and based on some logic I need to edit one of the column values.

     case  
               when (e."a" Like 'a%' Or e."b" Like 'b%') 
                And e."aa"='BW' And cast(e."abc" as decimal(10,4))=75.0 Then 'callitA'

               when (e."a" Like 'b%' Or e."b" Like 'a%') 
                And e."aa"='AW' And cast(e."abc" as decimal(10,4))=75.0 Then 'callitB'

else

'CallitC'

Answer

Ram Ghadiyaram picture Ram Ghadiyaram · Oct 11, 2016

Im not good in python. But will try to give some pointers of what I have done in scala.

Question : rdd.map and then do some logic checks. Is that the right approach?

Its one approach.

withColumn is another approach

DataFrame.withColumn method in pySpark supports adding a new column or replacing existing columns of the same name.

In this context you have to deal with Column via - spark udf or when otherwise syntax

for example :

from pyspark.sql import functions as F
df.select(df.name, F.when(df.age > 4, 1).when(df.age < 3, -1).otherwise(0)).show()


+-----+--------------------------------------------------------+
| name|CASE WHEN (age > 4) THEN 1 WHEN (age < 3) THEN -1 ELSE 0|
+-----+--------------------------------------------------------+
|Alice|                                                      -1|
|  Bob|                                                       1|
+-----+--------------------------------------------------------+


from pyspark.sql import functions as F
df.select(df.name, F.when(df.age > 3, 1).otherwise(0)).show()

+-----+---------------------------------+
| name|CASE WHEN (age > 3) THEN 1 ELSE 0|
+-----+---------------------------------+
|Alice|                                0|
|  Bob|                                1|
+-----+---------------------------------+

you can use udf instead of when otherwise as well.