對列中的唯一字符串進行分組並對單獨的列值執行 Function

Question

在我的 dataframe 中，我有“away_lineup”列，其中包含 5 個字符串的分組，還有一個“play_length”列，其中每行都有一個持續時間值。 我知道 np.unique 可以檢測唯一的字符串值並且 np.sum 值在列中添加值，但是我如何使用像 np.unique 這樣的 function 來檢測每個唯一的字符串並將字符串的“play_length”值相加連續發生？

away_lineup                                                                play_length
0  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons     0:00:05
1  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons     0:00:10
2  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons     0:00:20
3  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons     0:00:07
4  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons     0:00:25
5  Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, JJ Redick       0:00:14

我想要的 output 會是這樣的

player              play_length
Dario Saric             0:01:21
Robert Covington        0:01:21
Joel Embiid             0:01:21
Markelle Fultz          0:01:21
Ben Simmons             0:01:07
JJ Redick               0:00:14

從“away_lineup”中提取唯一名稱，存儲在新列“player”中，並且存在 player 值的行添加了它們的“play_length”值。

Answer 1

使用pandas.DataFrame.explode和pandas.to_timedelta ：

注意： pandas.DataFrame.explode適用於pandas >= 0.25

df['away_lineup'] = df['away_lineup'].str.split(', ')
df['play_length'] = pd.to_timedelta(df['play_length'])
new_df = df.explode('away_lineup').groupby('away_lineup').sum()
print(new_df)

Output：

                 play_length
away_lineup                 
Ben Simmons         00:01:07
Dario Saric         00:01:21
JJ Redick           00:00:14
Joel Embiid         00:01:21
Markelle Fultz      00:01:21
Robert Covington    00:01:21

Answer 2

如果您的 pandas 不支持explode ：

df['play_length'] = pd.to_timedelta(df['play_length'])

new_df = pd.concat((df[['play_length']], 
                    df['away_lineup'].str.split(',\s*', expand=True)), 
                   axis=1)

(new_df.melt(id_vars=['play_length'], 
             value_vars=new_df.columns[1:], 
             value_name='artist')
   .groupby('artist').play_length.sum()
)

Output：

artist
Ben Simmons        00:01:07
Dario Saric        00:01:21
JJ Redick          00:00:14
Joel Embiid        00:01:21
Markelle Fultz     00:01:21
Robert Covington   00:01:21
Name: play_length, dtype: timedelta64[ns]

Answer 3

檢查get_dummies的技巧

#df['play_length'] = pd.to_timedelta(df['play_length'])

df.away_lineup.str.get_dummies(',').mul(df.play_length,0).sum()
Out[372]: 
 Ben Simmons        00:01:07
 JJ Redick          00:00:14
 Joel Embiid        00:01:21
 Markelle Fultz     00:01:21
 Robert Covington   00:01:21
 Dario Saric        00:01:21
dtype: timedelta64[ns]

Answer 4

你可以像這樣使用explode和group by

import numpy as np
import pandas as pd

## create dummy data
arr = [("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons", "00:00:05"),
("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons", "00:00:10"),
("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons", "00:00:20"),
("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons", "00:00:07"),
("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, Ben Simmons", "00:00:25"),
("Dario Saric, Robert Covington, Joel Embiid, Markelle Fultz, JJ Redick", "00:00:14"),]

df = pd.DataFrame(arr, columns=["Player", "Play Time"])
df["Play Time"] = pd.to_timedelta(df["Play Time"])

## Solution
df["Player"] = df["Player"].str.split(",")
df.explode("Player").groupby("Player").sum()

output

            Play Time
Player  
Ben Simmons 00:01:07
JJ Redick   00:00:14
Joel Embiid 00:01:21
Markelle Fultz  00:01:21
Robert Covington    00:01:21
Dario Saric 00:01:21