将pyspark数组列乘以标量
我试图将数组类型的列乘以标量。此标量也是来自同一pyspark数据帧的值 例如,我有这样一个数据帧:将pyspark数组列乘以标量,pyspark,apache-spark-sql,Pyspark,Apache Spark Sql,我试图将数组类型的列乘以标量。此标量也是来自同一pyspark数据帧的值 例如,我有这样一个数据帧: df = sc.parallelize([([1, 2],3)]).toDF(["l","factor"]) +------+------+ | l|factor| +------+------+ |[1, 2]| 3| +------+------+ 我想要实现的是: +------+------+ | l|factor| +-
df = sc.parallelize([([1, 2],3)]).toDF(["l","factor"])
+------+------+
| l|factor|
+------+------+
|[1, 2]| 3|
+------+------+
我想要实现的是:
+------+------+
| l|factor|
+------+------+
|[3, 6]| 3|
+------+------+
这就是我尝试过的:
df.withColumn("l", lit("factor") * df.l)
但它返回一个类型不匹配错误。
如何将数组类型的列与数字相乘?来自
spark-2.4
使用变换
spark.sql(""" select l, factor, transform(l,x -> x * factor) as result from tmp """).show(10,False)
#+------+------+------+
#|l |factor|result|
#+------+------+------+
#|[1, 2]|3 |[3, 6]|
#+------+------+------+
df.withColumn("res",expr("""transform(l,x -> x*factor)""")).show()
#+------+------+------+
#| l|factor| res|
#+------+------+------+
#|[1, 2]| 3|[3, 6]|
#+------+------+------+
使用数据帧API:
spark.sql(""" select l, factor, transform(l,x -> x * factor) as result from tmp """).show(10,False)
#+------+------+------+
#|l |factor|result|
#+------+------+------+
#|[1, 2]|3 |[3, 6]|
#+------+------+------+
df.withColumn("res",expr("""transform(l,x -> x*factor)""")).show()
#+------+------+------+
#| l|factor| res|
#+------+------+------+
#|[1, 2]| 3|[3, 6]|
#+------+------+------+
正是我一直在寻找的!!!!谢谢!