如何将变量传递给UDAF(用户定义聚合函数)

时间:2020-09-21 12:25:44

标签: pyspark apache-spark-sql

import pandas as pd
import numpy as np
from pyspark.sql import SparkSession
import pyspark.sql.functions as F
from pyspark.sql.functions import PandasUDFType, pandas_udf
from pyspark.sql.types import *
import os



@pandas_udf(schema, functionType=PandasUDFType.GROUPED_MAP)
def split(df, validation_period):

   ""Logic""

    return df

def train_test_split(spark, data_frame, request_json_data):

    data_frame = spark.createDataFrame(data_frame)
    print(data_frame.schema)
 

    validation_period = request_json_data['validation_period']
    groupby_key = request_json_data['groupby_key']

    data_frame.groupby(groupby_key).apply(split, validation_period).show()

无法调用split函数,它给出错误。 apply()接受2个位置参数,但给出了3个。我想将validation_period作为参数传递给拆分函数。

1 个答案:

答案 0 :(得分:0)

简短的回答:您不能将额外的参数传递给熊猫分组地图udf,因为它仅将一个熊猫df作为参数。

长答案:还有其他方法可以将validation_period传递给函数

  1. 使用某种形式的闭包

    def split_fabric(validation_period):
        @pandas_udf(schema, functionType=PandasUDFType.GROUPED_MAP)
        def split(df):
    
            ""Logic""
    
            return df
    
  2. 将其作为列传递

    data_frame \
        .withColumn("validation_period", F.lit(validation_period)) \
        .groupby(groupby_key).apply(split, validation_period).show()