熊猫:按多列分组

时间:2020-04-27 12:09:23

标签: python pandas pandas-groupby

我正在学习pandas,并且具有很强的SQL背景,因此我需要重新考虑许多习惯和思维方式。虽然我认为我了解groupby()方法,但是我仍然无法弄清楚如何将其应用于多个列。

假设我们在数据库中有此表:

+----+--------------+-----------+--------------+-------+
| id | product_name | category  | subcategory  | price |
+----+--------------+-----------+--------------+-------+
|  1 | product1     | category1 | subcategory1 |  8.41 |
|  2 | product2     | category1 | subcategory1 | 62.74 |
|  3 | product3     | category1 | subcategory2 | 85.84 |
|  4 | product4     | category2 | subcategory2 | 32.71 |
|  5 | product5     | category2 | subcategory1 | 39.62 |
|  6 | product6     | category2 | subcategory1 | 37.43 |
|  7 | product7     | category3 | subcategory2 | 55.01 |
|  8 | product8     | category3 | subcategory1 | 26.91 |
|  9 | product9     | category3 | subcategory3 | 77.13 |
| 10 | product10    | category3 | subcategory3 | 40.79 |
+---+--------------+-----------+--------------+-------+

在多个列上进行汇总非常容易:

select category, subcategory, avg(price) as avg_price from my_table group by category, subcategory

返回以下内容:

+-----------+--------------+-----------+
| category  | subcategory  | avg_price |
+-----------+--------------+-----------+
| category1 | subcategory1 |    35.575 |
| category1 | subcategory2 |     85.84 |
| category2 | subcategory1 |    38.525 |
| category2 | subcategory2 |     32.71 |
| category3 | subcategory1 |     26.91 |
| category3 | subcategory2 |     55.01 |
| category3 | subcategory3 |     58.96 |
+-----------+--------------+-----------+

因此,以我显然不正确的理解,这在大熊猫中也是如此:

df['price'].groupby(df[['category', 'subcategory']]).mean()

返回ValueError: Grouper for '<class 'pandas.core.frame.DataFrame'>' not 1-dimensional,而:

 df['price'].groupby(df['category']).mean()

按预期工作。

有人可以帮我吗?

2 个答案:

答案 0 :(得分:1)

我认为您需要-

df.groupby(['category', 'subcategory'])['price'].mean()

答案 1 :(得分:0)

您必须修改groupby语法

df.groupby(['category', 'subcategory'])['price'].mean()