CODEWITHSUNDEEP
apache-sparkapache-spark-sql
Kora K
Kora K

Reputation: 435

Spark map dataframe to array

I'm working with the Spark MLlib PrefixSpan algorithm. I had some code working in Spark 1.6, but we recently moved to Spark 2.2.

I have a dataframe like this

viewsPurchasesGrouped: org.apache.spark.sql.DataFrame = [session_id: decimal(29,0), view_product_ids: array<bigint> ... 1 more field]

root
 |-- session_id: decimal(29,0) (nullable = true)
 |-- view_product_ids: array (nullable = true)
 |    |-- element: long (containsNull = true)
 |-- purchase_product_ids: array (nullable = true)
 |    |-- element: long (containsNull = true)

and in Spark 1.6, I used this piece of code to convert it to the appropriate dataframe for MLlib consumption:

import scala.collection.mutable.WrappedArray

val viewsPurchasesRddString = viewsPurchasesGrouped.map( row =>
  Array(
    Array(row.getAs[WrappedArray[String]](1).toArray), 
    Array(row.getAs[WrappedArray[String]](2).toArray)
  )
)

Since our switch, this doesn't work anymore.

I've tried this:

val viewsPurchasesRddString2 = viewsPurchasesGrouped.select("view_product_ids","purchase_product_ids").rdd.map( row =>
  Array(
    row.getSeq[Long](0).toArray, 
    row.getSeq[Long](1).toArray
  )
)

and see this baffling error message, which means that it took session_id and purchase_product_ids instead of view_product_ids and purchase_product_ids out of the original dataframe.

Job aborted due to stage failure: [...] scala.MatchError: [14545234113341303814564569524,WrappedArray(123, 234, 456, 678, 789)]

I've also tried this:

val viewsPurchasesRddString = viewsPurchasesGrouped.map {
   case Row(session_id: Long, view_product_ids: Array[Long], purchase_product_ids: Array[Long]) => 
     (view_product_ids, purchase_product_ids)
}

which fails with

viewsPurchasesRddString: org.apache.spark.sql.Dataset[(Array[Long], Array[Long])] = [_1: array<bigint>, _2: array<bigint>]
prefixSpan: org.apache.spark.mllib.fpm.PrefixSpan = org.apache.spark.mllib.fpm.PrefixSpan@10d69876
<console>:67: error: overloaded method value run with alternatives:
  [Item, Itemset <: Iterable[Item], Sequence <: Iterable[Itemset]](data: org.apache.spark.api.java.JavaRDD[Sequence])org.apache.spark.mllib.fpm.PrefixSpanModel[Item] <and>
  [Item](data: org.apache.spark.rdd.RDD[Array[Array[Item]]])(implicit evidence$1: 
scala.reflect.ClassTag[Item])org.apache.spark.mllib.fpm.PrefixSpanModel[Item] cannot be applied to (org.apache.spark.sql.Dataset[(Array[Long], Array[Long])])
   val model = prefixSpan.run(viewsPurchasesRddString)
                          ^

How do I port my code correctly?

Upvotes: 1

Views: 1886

Answers (1)

Raphael Roth
Raphael Roth

Reputation: 27373

your dataframe suggests that the columns are of type array<string>, so you should not access these using Seq[Long]. In spark 1.6, map on a dataframe automatically switched to RDD API, in Spark 2 you need to use rdd.map instead to do the same thing. So I would suggest this should work:

val viewsPurchasesRddString = viewsPurchasesGrouped.rdd.map( row =>
  Array(
    Array(row.getAs[WrappedArray[String]](1).toArray), 
    Array(row.getAs[WrappedArray[String]](2).toArray)
  )
)

Upvotes: 2

Related Questions