在第一次出现列值后删除所有行

Question

我的问题和这个一样，有一个额外的限制，我不知道如何解决：

Python PANDAS: Drop All Rows After First Occurrence of Column Value

在这篇文章中，问题是在第一次打开时删除所有行：

开始：

Rank Status
1    Closed
5    Closed
6    Open
9    Closed
10   Open

结果：

Rank Status
 1    Closed
 5    Closed
 6    Open

这是问题的最佳答案：

df = df.sort('Rank').reset_index()
df.loc[: df[(df['Status'] == 'Open')].index[0], :]

我有同样的问题，但我在同一个数据框中有多个商店，并希望为所有商店计算：

开始：

Shop Rank  Status
A    1     Closed
A    5     Closed
A    6     Open
A    9     Closed
A    10    Open
A    1     Closed
B    3     Closed
B    8     Closed
B    12    Open
B    15    Closed
...

我想要的结果：

Shop Rank  Status
A    1     Closed
A    5     Closed
A    6     Open
B    3     Closed
B    8     Closed
B    12    Open
...

我应该如何修改过去的答案以使其同时适应我的所有商店？

提前致谢

Answer 1

想法是通过比较Open 和Series.cummax 来处理每个组的列，因为还需要首先Open 是必要的转变：

mask = (df['Status'].eq('Open')
                    .groupby(df['Shop'])
                    .transform(lambda x: x.shift(fill_value=False).cummax()))

df = df[~mask]
print (df)
  Shop  Rank  Status
0    A     1  Closed
1    A     5  Closed
2    A     6    Open
6    B     3  Closed
7    B     8  Closed
8    B    12    Open

如果总是Open 每组可以使用：

def f(x):
    
    return x.loc[: x[(x['Status'] == 'Open')].index[0], :]

df = df.groupby('Shop', group_keys=False).apply(f)

---

def f(x):
    
    return x.loc[: (x['Status'] == 'Open').idxmax()]

df = df.groupby('Shop', group_keys=False).apply(f)

编辑：

使用一般数据进行测试：

print (df)
   Shop  Rank  Status
0     A     1  Closed
1     A     5  Closed
2     A     6    Open
3     A     9  Closed
4     A    10    Open
5     A     1  Closed
6     B     3  Closed
7     B     8  Closed
8     B    12    Open
9     B    15  Closed
10    C     8  Closed <- in C group no Open
11    C    12  Closed
12    D    15    Open <- in D group first Open
13    D    15  Closed

此解决方案失败，因为C 组中没有Open：

def f(x):
    
    return x.loc[: x[(x['Status'] == 'Open')].index[0], :]

df = df.groupby('Shop', group_keys=False).apply(f)

IndexError：索引 0 超出轴 0 的范围，大小为 0

---

def f(x):
    
    return x.loc[: (x['Status'] == 'Open').idxmax()]

df = df.groupby('Shop', group_keys=False).apply(f)

print (df)
   Shop  Rank  Status
0     A     1  Closed
1     A     5  Closed
2     A     6    Open
6     B     3  Closed
7     B     8  Closed
8     B    12    Open
10    C     8  Closed <- incorrect return first False row
12    D    15    Open

第一个解决方案运行良好：

mask = (df['Status'].eq('Open')
                    .groupby(df['Shop'])
                    .transform(lambda x: x.shift(fill_value=False).cummax()))

df1 = df[~mask]
print (df1)
   Shop  Rank  Status
0     A     1  Closed
1     A     5  Closed
2     A     6    Open
6     B     3  Closed
7     B     8  Closed
8     B    12    Open
10    C     8  Closed
11    C    12  Closed
12    D    15    Open