Python Spark DataFrame将元素转换为字符串

我有这样一个数据帧：

+---+--------------------+
|idn|     recommendations|
+---+--------------------+
|463|[[10955,0.0086656...|
|496|[[12767,0.0209305...|
|148|[[9813,0.00673213...|
|471|[[8537,0.00546676...|
|243|[[10846,0.0044064...|
|623|[[10955,0.3857911...|
|540|[[11463,0.0250675...|
|392|[[7177,0.01615425...|
|737|[[7994,0.12720428...|
|516|[[10955,0.4047550...|
+---+--------------------+

dataFrame.printSchema()


 root
 |-- idn: long (nullable = true)
 |-- recommendations: array (nullable = true)
 |    |-- element: struct (containsNull = true)
 |    |    |-- id_usn: long (nullable = true)
 |    |    |-- rating: double (nullable = true)

模式如下：

+---+--------------------+
|idn|     recommendations|
+---+--------------------+
|463|[[10955,0.0086656...|
|496|[[12767,0.0209305...|
|148|[[9813,0.00673213...|
|471|[[8537,0.00546676...|
|243|[[10846,0.0044064...|
|623|[[10955,0.3857911...|
|540|[[11463,0.0250675...|
|392|[[7177,0.01615425...|
|737|[[7994,0.12720428...|
|516|[[10955,0.4047550...|
+---+--------------------+

dataFrame.printSchema()


 root
 |-- idn: long (nullable = true)
 |-- recommendations: array (nullable = true)
 |    |-- element: struct (containsNull = true)
 |    |    |-- id_usn: long (nullable = true)
 |    |    |-- rating: double (nullable = true)

---

现在我想将列中的id\u usn和评级转换为字符串

您可以按如下方式强制转换嵌套结构列

col_schema = ArrayType(StructType([StructField('id_usn',StringType(),True),StructField('rating',StringType(),True)]))
df = dataFrame.select('idn',dataFrame.recommendations.cast(col_schema))
df.printSchema()

请试试这个，让我知道