一种做法是：透视，创建一个数组并对数组中的值求和

from pyspark.sql.functions import *

s =df.groupby('id').pivot('year').agg(sum('amount'))#Pivot
(s.withColumn('x', array(*[x for x in s.columns if x!='id']))#create array
         .withColumn('x', expr("reduce(x,cast(0 as bigint),(c,i)-> c+i)"))#sum
        ).show()

或使用pysparks内置聚合函数

s =df.groupby('id').pivot('year').agg(sum('amount'))#Pivot
(s.withColumn('x', array(*[x for x in s.columns if x!='id']))#create array
 .withColumn('x', expr("aggregate(x,cast(0 as bigint),(c,i)-> c+i)"))#sum
).show()