我希望能夠計算每個射手姓名的平均“目標”、“射門”和“失誤”，以用于進一步分析和可視化
下面的代碼為我提供了按“shooterName”排序的“event”列中的 3 個屬性（shot、goal、miss）的計數
資料框列：
season period time teamCode event goal xCord yCord xCordAdjusted yCordAdjusted ... playerPositionThatDidEvent timeSinceFaceoff playerNumThatDidEvent shooterPlayerId shooterName shooterLeftRight shooterTimeOnIce shooterTimeOnIceSinceFaceoff shotDistance
對應資料
2020 1 16 PHI SHOT 0 -74 29 74 -29 ... C 16 11 8478439.0 Travis Konecny R 16 16 32.649655
2020 1 34 PIT SHOT 0 49 -25 49 -25 ... C 34 9 8478542.0 Evan Rodrigues R 34 34 47.169906
2020 1 65 PHI SHOT 0 -52 -31 52 31 ... L 65 86 8480797.0 Joel Farabee L 31 31 48.270074
2020 1 171 PIT SHOT 0 43 39 43 39 ... C 42 9 8478542.0 Evan Rodrigues R 42 42 60.307545
2020 1 209 PHI MISS 0 -46 33 46 -33 ... D 38 5 8479026.0 Philippe Myers R 38 38 54.203321
當前代碼：
dft['count'] = df.groupby(['shooterName', 'event'])['event'].agg(['count'])
dft
電流輸出：
shooterName event count
A.J. Greer GOAL 1
MISS 6
SHOT 29
Aaron Downey GOAL 1
MISS 4
SHOT 35
Zenon Konopka GOAL 8
MISS 57
SHOT 176
期望輸出：
shooterName event count %totalshooterNameevents
A.J. Greer GOAL 1 .0277
MISS 6 .1666
SHOT 29 .805
Aaron Downey GOAL 1 .025
MISS 4 .1
SHOT 35 .875
Zenon Konopka GOAL 8 .0331
MISS 57 .236
SHOT 176 .7302
類似的東西。我的最終目標是能夠通過“shooterName”計算每個“事件”屬性占“事件”總數的百分比。下面我添加了一個列“%totalshooterNameevents”，它是“簡單的目標”、“射門”和“未命中”，由每個“射手名”的“目標、射門和未命中”之和計算
uj5u.com熱心網友回復：
更新
嘗試：
dft = df.groupby(['shooterName', 'event'])['event'].agg(['count']).reset_index()
dft['%total'] = dft.groupby('shooterName')['count'].apply(lambda x: x / sum(x))
print(dft)
# Output
shooterName event count %total
0 A.J. Greer GOAL 1 0.027778
1 A.J. Greer MISS 6 0.166667
2 A.J. Greer SHOT 29 0.805556
3 Aaron Downey GOAL 1 0.025000
4 Aaron Downey MISS 4 0.100000
5 Aaron Downey SHOT 35 0.875000
6 Zenon Konopka GOAL 8 0.033195
7 Zenon Konopka MISS 57 0.236515
8 Zenon Konopka SHOT 176 0.730290
沒有樣本，很難猜測你想要什么。嘗試：
import pandas as pd
import numpy as np
# Setup a Minimal Reproducible Example
np.random.seed(2021)
df = pd.DataFrame({'shooterName': np.random.choice(list('AB'), 20),
'event': np.random.choice(['shot', 'goal', 'miss'], 20)})
# Create an empty dataframe?
dft = pd.DataFrame(index=df['shooterName'].unique())
# Do stuff
grp = df.groupby('shooterName')
dft['count'] = grp.count()
dft = dft.join(grp['event'].value_counts().unstack('event')
.div(dft['count'], axis=0))
輸出：
>>> dft
count goal miss shot
A 12 0.416667 0.250 0.333333
B 8 0.500000 0.375 0.125000
轉載請註明出處，本文鏈接：https://www.uj5u.com/qianduan/397426.html
上一篇：室友打了一把端游，我入門了Vue