英文:
How to apply a function to the values of one column, then take the output and apply it to multiple other columns in pandas
问题
好的，以下是代码部分的翻译：
Alright y'all, maybe my strategy here isn't ideal, but I've got a very awkward dataset to work with and I need help.I have a pandas dataframe that's structured such that only the first column has values:df =|Ind| Column A | Column B | Column C || - | -------- | -------- | -------- || 0 | String1 | Null | Null || 1 | String2 | Null | Null |What I'd like to do is iteratively take the value from Column A and put it through a function whose output is a list. From there I need to fill the remaining columns with the output of the function, such that:df =|Ind| Column A | Column B | Column C || - | -------- | ---------------- | ---------------- || 0 | String1 | func(String1)[0] | func(String1)[1] || 1 | String2 | func(String2)[0] | func(String2)[1] |Thus far I've been trying to do this using anonymous functions, as such:df.iloc[:,1:].apply(lambda y: df["Column A"].apply(lambda x: list(map(func, x)))Which almost does what I want, but does not map the list into the respective columns, and the result is instead:df =|Ind| Column A | Column B | Column C || - | -------- | ------------- | ------------- || 0 | String1 | func(String1) | func(String1) || 1 | String2 | func(String2) | func(String2) |If there's a better approach I'm totally open.
英文:
Alright y'all, maybe my strategy here isn't ideal, but I've got a very awkward dataset to work with and I need help.
I have a pandas dataframe that's structured such that only the first column has values:
df =|Ind| Column A | Column B | Column C || - | -------- | -------- | -------- || 0 | String1 | Null | Null || 1 | String2 | Null | Null |
What I'd like to do is iteratively take the value from Column A and put it through a function whose output is a list. From there I need to fill the remaining columns with the output of the function, such that:
df =|Ind| Column A | Column B | Column C || - | -------- | ---------------- | ---------------- || 0 | String1 | func(String1)[0] | func(String1)[1] || 1 | String2 | func(String2)[0] | func(String2)[1] |
Thus far I've been trying to do this using anonymous functions, as such:
df.iloc[:,1:].apply(lambda y: df["Column A"].apply(lambda x: list(map(func, x)))
Which almost does what I want, but does not map the list into the respective columns, and the result is instead:
df =|Ind| Column A | Column B | Column C || - | -------- | ------------- | ------------- || 0 | String1 | func(String1) | func(String1) || 1 | String2 | func(String2) | func(String2) |
If there's a better approach I'm totally open.
答案1
得分: 0
你可以使用一个临时助手列与列表一起，然后在最后删除它，像这样：
df['temporary'] = df['Column A'].map(func)df['Column B'] = df['temporary'].str[0]df['Column C'] = df['temporary'].str[1]df = df.drop('temporary', axis=1)
英文:
You could use a temporary helper column with the list and then drop it at the end, like this:
df['temporary'] = df['Column A'].map(func)df['Column B'] = df['temporary'].str[0]df['Column C'] = df['temporary'].str[1]df = df.drop('temporary', axis=1)
答案2
得分: 0
函数式编程并不像它们所说的那么有趣。这里有一个过程化版本，它对列中的每个值应用一个函数，并扩展数据帧。请注意，该函数可以返回不同数量的结果。
import pandas as pd# 创建一个包含三行一列的数据帧df = pd.DataFrame(["foo", "bar", "fubar"], columns=["Column A"])# 应用于第1列并创建新行（可变长度是可以的）def fun(s):return list(c for c in s)# 通过应用函数创建一个新数据帧df2 = pd.DataFrame(fun(s) for s in df["Column A"])# 命名新列（第一列保持不变）column_names = {i: f"Column {'ABCDEFGHIJK'[i+1]}" for i in range(len(df2.columns))}# 使用新名称添加新列df = pd.concat([df, df2], axis=1).rename(columns=column_names)df
英文:
Functional programming is not as fun as they say it is. Here's a procedural version, that applies a function to each value in the column, and extends the data frame. Note that the function can return a variable number of results.
import pandas as pd# Make a three row, one column data framedf = pd.DataFrame(["foo","bar","fubar"],columns=["Column A"])# Apply to column 1 and create new rows (variable length is fine)def fun(s):return list(c for c in s)# Make a new data frame by applying functiondf2 = pd.DataFrame(fun(s) for s in df["Column A"])# Name new columns (first column remains same)column_names = {i:f"Column {'ABCDEFGHIJK'[i+1]}" for i in range(len(df2.columns))}# And add new columns using new namesdf = pd.concat([df,df2 ],axis=1).rename(columns=column_names)df
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论