Question

我正在读取具有以下架构的镶木地板文件：

df.printSchema()

root
 |-- time: integer (nullable = true)
 |-- amountRange: integer (nullable = true)
 |-- label: integer (nullable = true)
 |-- pcaVector: vector (nullable = true)

现在，我想测试Pyspark结构化的流，并且我想使用相同的镶木地板文件。我能够创建的最接近的架构是使用ArrayType，但是它不起作用：

schema = StructType(
    [
        StructField('time', IntegerType()),
        StructField('amountRange', IntegerType()),
        StructField('label', IntegerType()),
        StructField('pcaVector', ArrayType(FloatType()))

    ]
)
df_stream = spark.readStream\
    .format("parquet")\
    .schema(schema)\
    .load("/home/user/test_arch/data/fraud/")

Caused by: java.lang.ClassCastException: Expected instance of group converter but got "org.apache.spark.sql.execution.datasources.parquet.ParquetPrimitiveConverter"
        at org.apache.parquet.io.api.Converter.asGroupConverter(Converter.java:37)
        at org.apache.spark.sql.execution.datasources.parquet.ParquetRowConverter$RepeatedGroupConverter.<init>(ParquetRowConverter.scala:659)
        at org.apache.spark.sql.execution.datasources.parquet.ParquetRowConverter.org$apache$spark$sql$execution$datasources$parquet$ParquetRowConverter$$newConverter(ParquetRowConverter.scala:308)

如何使用VectorType创建一个仅在Scala中存在的Pyspark的StructType模式？

Answer 1

类型为VectorUDT

from pyspark.ml.linalg import VectorUDT

StructField('pcaVector', VectorUDT())

Pyspark模式中StructType的VectorType

1 个答案: