iiuc，这对你来说是个可行的解决方案-

在这里创建df

df = spark.createDataFrame([(1,[1.203, 3.455, 2.334, 1.2034, 0.345])],[ "col1","col2"])

# df.show(truncate=False)

df_grp = df.withColumn("explode_col", F.explode_outer("col2"))
df_grp = df_grp.groupBy("col1").pivot("explode_col").agg(F.avg("explode_col"))
df_grp.show()

输入和处理输出

+----+------------------------------------+
|col1|col2                                |
+----+------------------------------------+
|1   |[1.203, 3.455, 2.334, 1.2034, 0.345]|
+----+------------------------------------+

+----+-----+-----+------+-----+-----+
|col1|0.345|1.203|1.2034|2.334|3.455|
+----+-----+-----+------+-----+-----+
|   1|0.345|1.203|1.2034|2.334|3.455|
+----+-----+-----+------+-----+-----+

现在，您将观察到列名是一些意想不到的东西，#这里有一个小技巧来重新排序列名

count = 1
for col in df_grp.columns:
  if col != "col1":
    df_grp = df_grp.withColumnRenamed(col, "ftr"+str(count))
    print(col)
    print(count)
    count = count+1

最终输出

+----+-----+-----+------+-----+-----+
|col1| ftr1| ftr2|  ftr3| ftr4| ftr5|
+----+-----+-----+------+-----+-----+
|   1|0.345|1.203|1.2034|2.334|3.455|
+----+-----+-----+------+-----+-----+