Use hash partitioner

1 files changed, 11 insertions, 5 deletions

diff --git a/co-purchase-analysis/src/main/scala/Main.scala b/co-purchase-analysis/src/main/scala/Main.scala
index a12c850..0acce5b 100644
--- a/co-purchase-analysis/src/main/scala/Main.scala
+++ b/co-purchase-analysis/src/main/scala/Main.scala
@@ -1,5 +1,6 @@
 import org.apache.spark.sql.SparkSession
 import org.apache.spark.rdd.RDD
+import org.apache.spark.HashPartitioner
 import java.nio.file.{Paths, Files}
 
 /** A Spark application that analyzes co-purchased products.
@@ -106,18 +107,23 @@ object CoPurchaseAnalysis {
     *   RDD containing CoPurchase instances with purchase frequency counts
     */
   def processData(data: RDD[OrderProduct]): RDD[String] = {
-    data
+    val groups = data
       .groupBy(_.orderId)
+      .partitionBy(new HashPartitioner(200))
+      .persist()
+
+    val pairs = groups
       .flatMap { case (_, orders) =>
         orders.map(_.productId).toSeq.sorted.combinations(2).map {
           case List(p1, p2) =>
             ProductPair(Math.min(p1, p2), Math.max(p1, p2)) -> 1
         }
       }
-      .reduceByKey(_ + _)
-      .map { case (ProductPair(product1, product2), count) =>
-        s"${product1},${product2},${count}"
-      }
+    val coProducts = pairs.reduceByKey(_ + _)
+
+    coProducts.map { case (ProductPair(product1, product2), count) =>
+      s"${product1},${product2},${count}"
+    }
   }
 
   /** Main entry point for the application.