Clash Royale CLAN TAG#URR8PPP

spark aggregation for array column

I have a dataframe with a array column.

val json = """[
"id": 1, "value": [11, 12, 18],
"id": 2, "value": [23, 21, 29]
]"""

val df = spark.read.json(Seq(json).toDS)

scala> df.show
+---+------------+
| id| value|
+---+------------+
| 1|[11, 12, 18]|
| 2|[23, 21, 29]|
+---+------------+


Now I need to apply different aggregate functions to the value column.
I can call explode and groupBy, for example

explode

groupBy

df.select($"id", explode($"value").as("value")).groupBy($"id").agg(max("value"), avg("value")).show

+---+----------+------------------+
| id|max(value)| avg(value)|
+---+----------+------------------+
| 1| 18|13.666666666666666|
| 2| 29|24.333333333333332|
+---+----------+------------------+


What bothers me here is that I explode my DataFrame into a bigger one and then reduce it to the original calling groupBy.

groupBy

Is there a better (i.e. more efficient) way to call aggregated functions on array column? Probably I can implement UDF but I don't want to implement all aggregation UDFs myself.

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.