Scala Spark數據框-每行的總和為Array [Double]

Question

這是我的基本數據框：

root |-- user_id: string (nullable = true) 
     |-- review_id: string (nullable = true) 
     |-- review_influence: double (nullable = false)

目標是獲得每個user_id的review_influence之和。 所以我試圖匯總數據並總結如下：

val review_influence_listDF = review_with_influenceDF
.groupBy("user_id")
.agg(collect_list("review_id") as("list_review_id"), collect_list("review_influence") as ("list_review_influence"))
.agg(sum($"list_review_influence"))

但是我有這個錯誤：

org.apache.spark.sql.AnalysisException: cannot resolve 'sum(`list_review_influence`)' due to data type mismatch: function sum requires numeric types, not ArrayType(DoubleType,true);;

我該怎么辦？

Answer 1

您可以直接將agg函數中的列求和：

review_with_influenceDF
    .groupBy("user_id")
    .agg(collect_list($"review_id").as("list_review_id"), 
         sum($"review_influence").as("sum_review_influence"))