分箱極坐標

Question

我的問題很直接。 我想合並極坐標，這意味着我想要合並的域受 0 和 360 的限制，其中 0 = 360。由於數據的這種循環行為，這里開始我的問題，因為我想合並每 1 度從 0.5 度開始到 355.5 度（不幸的是，由於項目從 (0,1] 到 (359,360]) 分箱的性質，然后，我必須確保有一個從 (355.5) 開始的分箱,0.5]，這顯然不是默認情況下會發生的。

我編寫了這個腳本來更好地說明我在尋找什么：

bins_direction = np.linspace(0.5,360.5,360, endpoint = False)
points = np.random.rand(10000)*360
df = pd.DataFrame({'Points': points})
df['Bins'] = pd.cut(x= df['Points'],
                             bins=bins_direction)

您將看到，如果數據介於 355.5 和 0.5 度之間，則分箱將為 NaN。 我想找到一個解決方案，即 (355.5,0.5]

因此，我的結果（當然取決於您設置的種子）將如下所示：

          Points            Bins
0      17.102993    (16.5, 17.5]
1      97.665600    (97.5, 98.5]
2      46.697548    (46.5, 47.5]
3       9.832000     (9.5, 10.5]
4      21.260980    (20.5, 21.5]
5      47.433179    (46.5, 47.5]
6     359.813283             nan
7     355.654251  (355.5, 356.5]
8     0.23740105             nan

我希望它是：

          Points            Bins
0      17.102993    (16.5, 17.5]
1      97.665600    (97.5, 98.5]
2      46.697548    (46.5, 47.5]
3       9.832000     (9.5, 10.5]
4      21.260980    (20.5, 21.5]
5      47.433179    (46.5, 47.5]
6     359.813283    (359.5, 0.5]           
7     355.654251  (355.5, 356.5]
8     0.23740105    (359.5, 0.5]

Answer 1

由於您不能擁有(355.5, 0.5]形式的pandas區間，因此您只能將它們作為字符串：

bins = [0] + list(np.linspace(0.5,355.5,356)) + [360]

df = pd.DataFrame({'Points': [0,1,350,356, 357, 359]})

(pd.cut(df['Points'], bins=bins, include_lowest=True)
   .astype(str)
   .replace({'(-0.001, 0.5]':'(355.5,0.5]', '(355.5, 360.0]':'(355.5,0.5]'})
)

輸出：

0       (355.5,0.5]
1        (0.5, 1.5]
2    (349.5, 350.5]
3       (355.5,0.5]
4       (355.5,0.5]
5       (355.5,0.5]
Name: Points, dtype: object