使用 Numpy 向量化的循环

Question

我正在尝试复制使用循环和 numpy 向量化的结果，在此处找到文章 (https://towardsdatascience.com/how-to-make-your-pandas-loop-71-803-times-faster-805030df4f06)。这篇文章不包含来自运行的数据或结果，但我能够在网上找到数据。我想为自己的工作复制结果，但输出不正确。

我已经包含了文章中的一小部分原始数据框和相应的代码：

import pandas as pd 
data = {'HomeTeam':['Burnley','Crystal Palace','Everton','Hull','Man City','Middlesbrough','Southampton',
'Arsenal','Bournemouth','Chelsea','Man United','Burnley','Leicester','Stoke'], 'AwayTeam':['Swansea','West Brom','Tottenham','Leicester','Sunderland','Stoke','Watford','Liverpool','Man United',
'West Ham','Southampton','Liverpool','Arsenal','Man City'], 'FTR': ['A','A','D','H','H','D','D','A','A','H','H','H','D','A']} 

leaguedf = pd.DataFrame(data) 

def soc_iter(TEAM,home,away,ftr):
    leaguedf['Draws'] = 'No_Game'
    leaguedf.loc[((home == TEAM) & (ftr == 'D')) | ((away == TEAM) & (ftr == 'D')), 'Draws'] = 'Draw'
    leaguedf.loc[((home == TEAM) & (ftr != 'D')) | ((away == TEAM) & (ftr != 'D')), 'Draws'] = 'No_Draw'

leaguedf['Draws']=soc_iter('Arsenal',leaguedf['HomeTeam'].values, leaguedf['AwayTeam'].values, leaguedf['FTR'].values)
leaguedf

当我运行代码输出列 'Draws' 只生成 'None' 的输出，而不是 'Draw' 或 'No_Draw'。

代码有什么问题？

Answer 1

你的函数没有 return 任何东西，所以 Draws 将全部 None，你不必分配任何东西，函数内的代码已经在创建 Draw 列：

import pandas as pd 
data = {'HomeTeam':['Burnley','Crystal Palace','Everton','Hull','Man City','Middlesbrough','Southampton',
'Arsenal','Bournemouth','Chelsea','Man United','Burnley','Leicester','Stoke'], 'AwayTeam':['Swansea','West Brom','Tottenham','Leicester','Sunderland','Stoke','Watford','Liverpool','Man United',
'West Ham','Southampton','Liverpool','Arsenal','Man City'], 'FTR': ['A','A','D','H','H','D','D','A','A','H','H','H','D','A']} 

leaguedf = pd.DataFrame(data) 

def soc_iter(TEAM,home,away,ftr):
    leaguedf['Draws'] = 'No_Game'
    leaguedf.loc[((home == TEAM) & (ftr == 'D')) | ((away == TEAM) & (ftr == 'D')), 'Draws'] = 'Draw'
    leaguedf.loc[((home == TEAM) & (ftr != 'D')) | ((away == TEAM) & (ftr != 'D')), 'Draws'] = 'No_Draw'

soc_iter('Arsenal',leaguedf['HomeTeam'].values, leaguedf['AwayTeam'].values, leaguedf['FTR'].values)
leaguedf

           HomeTeam     AwayTeam FTR    Draws
0          Burnley      Swansea   A  No_Game
1   Crystal Palace    West Brom   A  No_Game
2          Everton    Tottenham   D  No_Game
3             Hull    Leicester   H  No_Game
4         Man City   Sunderland   H  No_Game
5    Middlesbrough        Stoke   D  No_Game
6      Southampton      Watford   D  No_Game
7          Arsenal    Liverpool   A  No_Draw
8      Bournemouth   Man United   A  No_Game
9          Chelsea     West Ham   H  No_Game
10      Man United  Southampton   H  No_Game
11         Burnley    Liverpool   H  No_Game
12       Leicester      Arsenal   D     Draw
13           Stoke     Man City   A  No_Game

使用 Numpy 向量化的循环

Loops using Numpy Vectorization

numpy

vectorization