np.argsort 如何在 Pandas DataFrame 中工作？

How np.argsort works in pandas DataFrame?

提问by Karthik Elangovan

I have the following pandas DataFrame named index:

我有以下 Pandas DataFrame 命名index：

tz
                       521.0
Africa/Cairo             3.0
Africa/Casablanca        1.0   
Africa/Ceuta             2.0
Africa/Johannesburg      1.0
dtype: float64 

when I apply index.argsort()I get something like this:

当我申请时，index.argsort()我得到这样的信息：

tz
                       2
Africa/Cairo           4
Africa/Casablanca      3
Africa/Ceuta           1
Africa/Johannesburg    0
dtype: int64

Can someone explain to me where the numbers: 2,4,3,1,0come? I know they're index range from 0 to 4 but I can't find any logic in their order.

有人可以向我解释数字在哪里：2,4,3,1,0来吗？我知道它们的索引范围从 0 到 4，但我找不到它们的顺序中的任何逻辑。

回答by piRSquared

argsortreturns the index positions of the values being sorted if they were to be sorted. Keep in mind that this is a numpy function and its assignment to series or dataframe indices is erroneous.

argsort如果要排序，则返回正在排序的值的索引位置。请记住，这是一个 numpy 函数，它对系列或数据帧索引的分配是错误的。

• 2refers to the item in the 2position (3rd) was the minimum
  • this was 1.0
• 4refers to the item in the 4position (5th) was next
  • also 1.0
• 3(4th position) was a 2.0
• 1(2nd position) was a 3.0
• 0(1st position) was a 521.0and the maximum

• 2指2位置 (3rd) 中的项目是最小的
  • 这是 1.0
• 4指4位置（第 5 个）中的项目是下一个
  • 还 1.0
• 3（第 4 位）是一个 2.0
• 1（第二个位置）是一个 3.0
• 0（第一个位置）是一个521.0和最大值

---

---

It's more appropriate to assign to an array and use as a slice

分配给数组并用作切片更合适

a = s.values.argsort()
s.iloc[a]

tz
Africa/Casablanca        1.0
Africa/Johannesburg      1.0
Africa/Ceuta             2.0
Africa/Cairo             3.0
                       521.0
Name: value, dtype: float64