如何从数据框列中提取特定项并将其用作其余项的标签？

Question

我有一个只有一列的数据框，我想提取某些项目，并将它们转换成一个单独的列，用作其他项目的标签。这有点难以解释，例如，如果我有：

pd.DataFrame({'Fruits': ['Apple', 'Gala', 'Fuji', 'Grannysmith', 'Honeycrisp', 'Golden', 'pink', 'Orange', 'blood orange', 'Mandrin', 'Tangerine', 'Clementine', 'Banana', 'baby', 'manzano', 'burro']})

          Fruits
0          Apple
1           Gala
2           Fuji
3    Grannysmith
4     Honeycrisp
5         Golden
6           pink
7         Orange
8   blood orange
9        Mandrin
10     Tangerine
11    Clementine
12        Banana
13          baby
14       manzano
15         burro

但我想把它转换成：

    Fruits  Types
0   Apple   Gala
1   Apple   Fuji
2   Apple   Grannysmith
3   Apple   Honeycrisp
4   Apple   Golden
5   Apple   pink
6   Orange  blood orange
7   Orange  Mandrin
8   Orange  Tangerine
9   Orange  Clementine
10  Banana  baby
11  Banana  manzano
12  Banana  burro

如何将第一个数据帧转换为第二个数据帧？我对此感到困惑，尤其是当水果种类繁多，种类繁多的时候。你知道吗

Answer 1

我会用一些标准逻辑构建一个字典映射，然后将其用于操作

fruit_classes = ['Apple', 'Orange', 'Banana']

last_class = None
fruit_map = {}

for fruit in df.Fruits:
  if fruit in fruit_classes:
    last_class = fruit
  elif last_class is not None:
    fruit_map[fruit] = last_class

df.assign(Types=df.Fruits, Fruits=df.Fruits.map(fruit_map)).dropna()

    Fruits         Types
1    Apple          Gala
2    Apple          Fuji
3    Apple   Grannysmith
4    Apple    Honeycrisp
5    Apple        Golden
6    Apple          pink
8   Orange  blood orange
9   Orange       Mandrin
10  Orange     Tangerine
11  Orange    Clementine
13  Banana          baby
14  Banana       manzano
15  Banana         burro

---

或者理解

fruit_classes = ['Apple', 'Orange', 'Banana']

水果类=['苹果'，'橘子'，'香蕉']

pd.DataFrame(
    [[x, None][::(x in fruit_classes) * 2 - 1] for x in df.Fruits],
    columns=['Fruits', 'Types']
).assign(Fruits=lambda d: d.Fruits.ffill()).dropna()

    Fruits         Types
1    Apple          Gala
2    Apple          Fuji
3    Apple   Grannysmith
4    Apple    Honeycrisp
5    Apple        Golden
6    Apple          pink
8   Orange  blood orange
9   Orange       Mandrin
10  Orange     Tangerine
11  Orange    Clementine
13  Banana          baby
14  Banana       manzano
15  Banana         burro

Answer 2

首先需要定义列表中的结果，然后创建一个新列，用^{}重复结果，用^{}替换缺失值和正向填充，然后用^{}删除两列中的相同值，最后设置新列名称：

L = ['Apple','Orange','Banana']

df['a'] = df['Fruits'].where(df['Fruits'].isin(L)).ffill()
df = df.loc[df['a'] != df['Fruits'], ['a','Fruits']]
df.columns = ['Fruits','Types']
print (df)

    Fruits         Types
1    Apple          Gala
2    Apple          Fuji
3    Apple   Grannysmith
4    Apple    Honeycrisp
5    Apple        Golden
6    Apple          pink
8   Orange  blood orange
9   Orange       Mandrin
10  Orange     Tangerine
11  Orange    Clementine
13  Banana          baby
14  Banana       manzano
15  Banana         burro