合并 pandas 数据框中的某些列并复制其他列

Question

我有一个与如何从极坐标数据框中的列表中选择所有列类似的问题，但略有不同：

import polars as pl
import numpy as np
import string

n = 1000
letters = list(string.ascii_letters)
uppercase = list(string.ascii_uppercase)
words, groups = [], []
for i in range(n):
    word = ''.join([np.random.choice(letters) for _ in range(np.random.randint(3, 20))])
    words.append(word)
    group = np.random.choice(uppercase)
    groups.append(group)
df = pl.DataFrame(
    {
        "a_0": np.linspace(0, 1, n),
        "a_1": np.linspace(1, 2, n),
        "a_2": np.linspace(2, 3, n),
        "a_3": np.linspace(3, 4, n),
        "a_4": np.linspace(4, 5, n),
        "a_5": np.linspace(5, 6, n),
        "a_6": np.linspace(6, 7, n),
        "b_0": np.random.rand(n),
        "b_1": 2*np.random.rand(n),
        "b_2": 3*np.random.rand(n),
        "b_3": 4*np.random.rand(n),
        "b_4": 5*np.random.rand(n),
        "b_5": 6*np.random.rand(n),
        "b_6": 7*np.random.rand(n),
        "words": words,
        "groups": groups,
    }
)

我想再次将列

a_0

、

a_1

...连接成列

a

，将列

b_0

、

b_1

...连接成列

b

。然而，与之前的问题不同的是，这次

a = [a_0;  a_1; ...]

。即，首先是

a_0

的所有元素，然后是

a_1

的所有元素，依此类推。名称不以

_

结尾的所有列，后跟一个数字（在本例中为

words

）和

groups

）必须重复足够多的次数以匹配

a

的长度。我怎样才能做到这一点？

Answer 1

尝试下面的代码，它将连接列

import polars as pl
import numpy as np
import string

n = 1000
letters = list(string.ascii_letters)
uppercase = list(string.ascii_uppercase)
words, groups = [], []
for i in range(n):
    word = ''.join([np.random.choice(letters) for _ in range(np.random.randint(3, 20))])
    words.append(word)
    group = np.random.choice(uppercase)
    groups.append(group)
df = pl.DataFrame(
    {
        "a_0": np.linspace(0, 1, n),
        "a_1": np.linspace(1, 2, n),
        "a_2": np.linspace(2, 3, n),
        "a_3": np.linspace(3, 4, n),
        "a_4": np.linspace(4, 5, n),
        "a_5": np.linspace(5, 6, n),
        "a_6": np.linspace(6, 7, n),
        "b_0": np.random.rand(n),
        "b_1": 2*np.random.rand(n),
        "b_2": 3*np.random.rand(n),
        "b_3": 4*np.random.rand(n),
        "b_4": 5*np.random.rand(n),
        "b_5": 6*np.random.rand(n),
        "b_6": 7*np.random.rand(n),
        "words": words,
        "groups": groups,
    }
)

# Select columns with names ending with a_ followed by a digit
a_columns = [col for col in df.columns if col.endswith("_") and col[-2:].isdigit()]

# Select columns with names not ending with a_ followed by a digit
other_columns = [col for col in df.columns if not col.endswith("_") or not col[-2:].isdigit()]

# Concatenate the selected columns into a single column
concatenated_a = df.select(a_columns).hstack()

# Duplicate other columns to match the length of concatenated_a
duplicated_other_columns = df.select(other_columns).hstack().clone().extend(concatenated_a.len())

# Join the concatenated column with duplicated other columns
result = concatenated_a.join(duplicated_other_columns)

print(result)