pandas对具有多个条目的行进行get_dummies操作

huangapple
go评论92阅读模式
英文:

pandas get_dummies on rows with multiple entries

问题

如果我有这样的数据框：

Fruits
apple, banana, strawberry
apple
strawberry, apple

我在创建像这样的虚拟列时遇到了问题，因为每一行可能有多种水果。这将是我期望的结果：

apple banana strawberry
1 1 1
1 0 0
1 0 1

尝试单独使用get_dummies函数不起作用，因为它会创建以下列：

apple,banana,strawberry apple strawberry,apple
1 0 0
0 1 0
0 0 1

感谢任何帮助，谢谢！

英文:

If I have a dataframe like this:

Fruits
apple, banana, strawberry
apple
strawberry, apple

I am having trouble creating dummy columns for something like this, as it may have multiple fruits in each row. This would be my desired outcome:

apple banana strawberry
1 1 1
1 0 0
1 0 1

Trying the get_dummies function by itself does not work, since it will create the columns like:

apple,banana,strawberry apple strawberry,apple
1 0 0
0 1 0
0 0 1

Any help is appreciated, thank you!

答案1

得分: 2

以下是您要翻译的代码部分：

  1. import pandas as pd
  2. data = [
  3.     ['apple, banana, strawberry'],
  4.     ['apple'],
  5.     ['strawberry, apple']
  6. ]
  8. df = pd.DataFrame(data, columns=['Fruits'])
  9. print(df)
  11. columns = set()
  12. for row in df['Fruits'].to_list():
  13.     columns |= set(row.split(', '))
  15. rows = []
  16. for row in df['Fruits'].to_list():
  17.     rows.append([int(c in row) for c in columns])
  19. columns = list(columns)
  20. df = pd.DataFrame(rows, columns=columns)
  21. print(df)

Output:

  1.                       Fruits
  2. 0  apple, banana, strawberry
  3. 1                      apple
  4. 2          strawberry, apple
  5.    apple  banana  strawberry
  6. 0      1       1           1
  7. 1      1       0           0
  8. 2      1       0           1
英文:

Here's one way, using the technique I mentioned in my comment:

  1. import pandas as pd
  2. data = [
  3.     ['apple, banana, strawberry'],
  4.     ['apple'],
  5.     ['strawberry, apple']
  6. ]
  8. df = pd.DataFrame(data, columns=['Fruits'])
  9. print(df)
  11. columns = set()
  12. for row in df['Fruits'].to_list():
  13.     columns |= set( row.split(', ') )
  15. rows = []
  16. for row in df['Fruits'].to_list():
  17.     rows.append( [int(c in row) for c in columns] )
  19. columns = list(columns)
  20. df = pd.DataFrame(rows, columns=columns)
  21. print(df)

Output:

  1.                       Fruits
  2. 0  apple, banana, strawberry
  3. 1                      apple
  4. 2          strawberry, apple
  5.    apple  banana  strawberry
  6. 0      1       1           1
  7. 1      1       0           0
  8. 2      1       0           1

答案2

得分: 1

以下是翻译好的部分：

  1. dummy_df = df["Fruits"].str.get_dummies(", ")

输出：

  1.       apple banana strawberry
  2.     0     1      1          1
  3.     1     1      0          0
  4.     2     1      0          1
英文:

You can just do this

  1. dummy_df = df["Fruits"].str.get_dummies(", ")

Output

  1.   apple banana strawberry
  2. 0     1      1          1
  3. 1     1      0          0
  4. 2     1      0          1

答案3

得分: 0

你还可以使用标准库来获得整洁且漂亮的代码。

  1. from itertools import chain
  3. def create_dummy(df: pd.DataFrame) -> pd.DataFrame:
  4.     # 获取所有唯一水果的列表
  5.     fruits_series = df["Fruits"].str.split(", ")
  6.     unique_fruits = set(chain.from_iterable(fruits_series))
  8.     # 创建虚拟DataFrame
  9.     tdf = pd.DataFrame(0,
  10.                        columns=list(unique_fruits),
  11.                        index=df.index,
  12.                        dtype=pd.UInt8Dtype)
  14.     # 更新标签
  15.     for ind, targets in zip(df.index, fruits_series):
  16.         tdf.loc[ind, targets] = 1
  18.     return tdf
  20. 用法：
  22. ```python
  23. df = pd.DataFrame(["apple, orange, grape", 
  24.                    "apple", 
  25.                    "banana, strawberry"], columns=["Fruits"])
  26. print(create_dummy(df))

输出：

  1.        grape  strawberry  orange  apple  banana
  2. 0         1           0       1      1       0
  3. 1         0           0       0      1       0
  4. 2         0           1       0      0       1

已在 python==3.10.8 和 pandas==1.5.2 上测试过。

英文:

You can also do so with a help of a standard library to get neat and pretty code.

  1. from itertools import chain
  3. def create_dummy(df: pd.DataFrame) -> pd.DataFrame:
  4.     # Get list of all unique fruits
  5.     fruits_series = df["Fruits"].str.split(", ")
  6.     unique_fruits = set(chain.from_iterable(fruits_series))
  8.     # Create dummy DataFrame
  9.     tdf = pd.DataFrame(0,
  10.                        columns=list(unique_fruits),
  11.                        index=df.index,
  12.                        dtype=pd.UInt8Dtype)
  14.     # Update labels
  15.     for ind, targets in zip(df.index, fruits_series):
  16.         tdf.loc[ind, targets] = 1
  18.     return tdf

Usage:

  1. df = pd.DataFrame(["apple, orange, grape", 
  2.                    "apple", 
  3.                    "banana, strawberry"], columns=["Fruits"])
  4. print(create_dummy(df))

Output:

  1.        grape  strawberry  orange apple banana
  2.  0      1           0       1     1      0
  3.  1      0           0       0     1      0
  4.  2      0           1       0     0      1

Tested on python==3.10.8 and pandas==1.5.2.

huangapple
  • 本文由 发表于 2023年7月28日 06:24:27
  • 转载请务必保留本文链接：https://go.coder-hub.com/76783764.html
  • dataframe
  • pandas
  • python
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定