比较 pandas.DataFrame 中列之间的分类变量
Comparing categorical variables between columns in pandas.DataFrame
如何使用设置的分类规则而不是词典顺序规则进行比较？
给定数据集：
df = pd.DataFrame({
'NUMBER':[12, 26, 16, 34, 38, 1, 26, 8],
'SHIRT_SIZE':['S', 'M', 'XL', 'L', 'S', 'M', 'L', 'XL'],
'SHIRT_SIZE2':['M', 'S', 'L', 'XL', 'M', 'L', 'XL', 'S']
})
from pandas.api.types import CategoricalDtype
c_dtype = CategoricalDtype(categories = ["S","M","L","XL"],ordered = True)
df['SHIRT_SIZE'] = df['SHIRT_SIZE'].astype(c_dtype)
df['SHIRT_SIZE2'] = df['SHIRT_SIZE2'].astype(c_dtype)
NUMBER
SHIRT_SIZE
SHIRT_SIZE2
0
12
S
M
1
26
M
S
2
16
XL
L
3
34
L
XL
4
38
S
M
5
1
M
L
6
26
L
XL
7
8
XL
S
'SHIRT_SIZE'和'SHIRT_SIZE2'的dtype是Categories (4, object): ['S' < 'M' < 'L' < 'XL']
我想比较两列 'SHIRT_SIZE' 和 'SHIRT_SIZE2'
之间的衬衫尺码
我尝试过：
def compare_size(row):
if (row['SHIRT_SIZE'] < row['SHIRT_SIZE2']):
return 'SMALLER'
elif (row['SHIRT_SIZE'] > row['SHIRT_SIZE2']):
return 'LARGER'
else:
return 'SAME'
df['COMPARE_SIZE'] = df.apply(lambda row: compare_size(row), axis=1)
导致：
NUMBER
SHIRT_SIZE
SHIRT_SIZE2
COMPARE_SIZE
0
12
S
M
LARGER
1
26
M
S
SMALLER
2
16
XL
L
LARGER
3
34
L
XL
SMALLER
4
38
S
M
LARGER
5
1
M
L
LARGER
6
26
L
XL
SMALLER
7
8
XL
S
LARGER
请注意，有一些行 例如第 0 行 'S' -> 'M' 和第 1 行 'M' -> 'S' 不遵循我们的分类 dtype 规则的顺序
从逻辑上讲，解释是：“SHIRT_SIZE 比 SHIRT_SIZE2”
我猜测字符串的词典顺序是用于比较这些衬衫尺码的基本规则，而不是我们在 Categories (4, object): ['S' < 'M' < 'L' < 'XL'].
中设置的分类规则
我希望按照分类顺序比较衬衫尺码。
使用 numpy select 比较值并生成新列：
condlist = [df.SHIRT_SIZE.gt(df.SHIRT_SIZE2), df.SHIRT_SIZE.lt(df.SHIRT_SIZE2)]
result_list = ["LARGER", "SMALLER"]
compare_size = np.select(condlist, result_list, "SAME")
df.assign(compare_size=compare_size)
NUMBER SHIRT_SIZE SHIRT_SIZE2 compare_size
0 12 S M SMALLER
1 26 M S LARGER
2 16 XL L LARGER
3 34 L XL SMALLER
4 38 S M SMALLER
5 1 M L SMALLER
6 26 L XL SMALLER
7 8 XL S LARGER
如何使用设置的分类规则而不是词典顺序规则进行比较？
给定数据集：
df = pd.DataFrame({
'NUMBER':[12, 26, 16, 34, 38, 1, 26, 8],
'SHIRT_SIZE':['S', 'M', 'XL', 'L', 'S', 'M', 'L', 'XL'],
'SHIRT_SIZE2':['M', 'S', 'L', 'XL', 'M', 'L', 'XL', 'S']
})
from pandas.api.types import CategoricalDtype
c_dtype = CategoricalDtype(categories = ["S","M","L","XL"],ordered = True)
df['SHIRT_SIZE'] = df['SHIRT_SIZE'].astype(c_dtype)
df['SHIRT_SIZE2'] = df['SHIRT_SIZE2'].astype(c_dtype)
| NUMBER | SHIRT_SIZE | SHIRT_SIZE2 | |
|---|---|---|---|
| 0 | 12 | S | M |
| 1 | 26 | M | S |
| 2 | 16 | XL | L |
| 3 | 34 | L | XL |
| 4 | 38 | S | M |
| 5 | 1 | M | L |
| 6 | 26 | L | XL |
| 7 | 8 | XL | S |
'SHIRT_SIZE'和'SHIRT_SIZE2'的dtype是Categories (4, object): ['S' < 'M' < 'L' < 'XL']
我想比较两列 'SHIRT_SIZE' 和 'SHIRT_SIZE2'
我尝试过：
def compare_size(row):
if (row['SHIRT_SIZE'] < row['SHIRT_SIZE2']):
return 'SMALLER'
elif (row['SHIRT_SIZE'] > row['SHIRT_SIZE2']):
return 'LARGER'
else:
return 'SAME'
df['COMPARE_SIZE'] = df.apply(lambda row: compare_size(row), axis=1)
导致：
| NUMBER | SHIRT_SIZE | SHIRT_SIZE2 | COMPARE_SIZE | |
|---|---|---|---|---|
| 0 | 12 | S | M | LARGER |
| 1 | 26 | M | S | SMALLER |
| 2 | 16 | XL | L | LARGER |
| 3 | 34 | L | XL | SMALLER |
| 4 | 38 | S | M | LARGER |
| 5 | 1 | M | L | LARGER |
| 6 | 26 | L | XL | SMALLER |
| 7 | 8 | XL | S | LARGER |
请注意，有一些行 例如第 0 行 'S' -> 'M' 和第 1 行 'M' -> 'S' 不遵循我们的分类 dtype 规则的顺序
从逻辑上讲，解释是：“SHIRT_SIZE
我猜测字符串的词典顺序是用于比较这些衬衫尺码的基本规则，而不是我们在 Categories (4, object): ['S' < 'M' < 'L' < 'XL'].
我希望按照分类顺序比较衬衫尺码。
使用 numpy select 比较值并生成新列：
condlist = [df.SHIRT_SIZE.gt(df.SHIRT_SIZE2), df.SHIRT_SIZE.lt(df.SHIRT_SIZE2)]
result_list = ["LARGER", "SMALLER"]
compare_size = np.select(condlist, result_list, "SAME")
df.assign(compare_size=compare_size)
NUMBER SHIRT_SIZE SHIRT_SIZE2 compare_size
0 12 S M SMALLER
1 26 M S LARGER
2 16 XL L LARGER
3 34 L XL SMALLER
4 38 S M SMALLER
5 1 M L SMALLER
6 26 L XL SMALLER
7 8 XL S LARGER