withColumn不允许我使用max()函数生成新列 [英] withColumn not allowing me to use max() function to generate a new column

问题描述

我有一个像这样的数据集:

I have a dataset like this:

a = sc.parallelize([[1,2,3],[0,2,1],[9,8,7]]).toDF(["one", "two", "three"])

我想拥有一个数据集，该数据集会添加一个新列，该列等于其他三列中的最大值. 输出看起来像这样:

I want to have a dataset that adds a new column that is equal to the largest value in the other three columns. The output would look like this:

+----+----+-----+-------+
|one |two |three|max_col|
+----+----+-----+-------+
|   1|   2|    3|      3|
|   0|   2|    1|      2|
|   9|   8|    7|      9|
+----+----+-----+-------+

我想我会使用withColumn，就像这样:

I thought I would use withColumn, like so:

b = a.withColumn("max_col", max(a["one"], a["two"], a["three"]))

但这会产生错误

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/spark152/python/pyspark/sql/column.py", line 418, in __nonzero__
    raise ValueError("Cannot convert column into bool: please use '&' for 'and', '|' for 'or', "
ValueError: Cannot convert column into bool: please use '&' for 'and', '|' for 'or', '~' for 'not' when building DataFrame boolean expressions.

奇怪. max返回bool吗?不符合 max 上的文档.好的.很奇怪.

Odd. Does max return a bool? Not according to the documentation on max. Okay. Weird.

我觉得这很奇怪:

b = a.withColumn("max_col", a["one"] + a["two"] + a["three"]))

它起作用的事实使我更加强烈地认为max表现出某种我不理解的方式.

And the fact that it works makes me think even more strongly that max is behaving some way I don't understand.

我也尝试过b = a.withColumn("max_col", max([a["one"], a["two"], a["three"]]))，它将三列作为列表而不是3个separte元素传递.这会产生与上述相同的错误.

I also tried b = a.withColumn("max_col", max([a["one"], a["two"], a["three"]])), which passes in the three columns as a list rather than 3 separte elements. This yields the same error as above.

推荐答案

实际上，您需要的是greatest而不是max:

Actually what you need here is greatest not max:

from pyspark.sql.functions import greatest

a.withColumn("max_col", greatest(a["one"], a["two"], a["three"]))

为了完整起见，您可以使用least查找最小值:

And just for completeness you can use least to find the minimum:

from pyspark.sql.functions import least

a.withColumn("min_col", least(a["one"], a["two"], a["three"]))

关于错误，您看到它非常简单. max取决于丰富的比较.比较两列时，您会得到Column:

Regarding the error you see it is quite simple. max depends on the rich comparisons. When you compare two columns you get a Column:

type(col("a") < col("b")
## pyspark.sql.column.Column

PySpark明确禁止将列转换为布尔值(您可以检查Column.__nonzero__源代码)，因为它根本没有意义.它只是一个逻辑表达式，无法在驱动程序上下文中进行评估.

PySpark explicitly forbids converting columns to booleans (you can check Column.__nonzero__ source) because it is simply meaningless. It is only a logical expression which cannot be evaluated in the driver context.

这篇关于withColumn不允许我使用max()函数生成新列的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！