使用 Python 模糊匹配同一数据框中的两列

Question

我在同一个数据框中有两个数据集，每个数据集都显示一个公司列表。一个数据集来自 2017 年，另一个数据集来自今年。我正在尝试将两个公司的数据集相互匹配，并认为模糊匹配 (FuzzyWuzzy) 是执行此操作的最佳方法。使用部分比率，我想简单地列出具有如下值的列：去年公司名称，最高模糊匹配率，今年公司与最高分相关联。原始数据框已被赋予变量 "data"，去年的公司名称位于 "Company" 列下，今年的公司名称位于 "Company name" 列下。为完成此任务，我尝试使用 extractOne 模糊匹配过程创建一个函数，然后将该函数应用于数据框中的每个 value/row。然后我会将结果添加到我的原始数据框中。

下面是代码：

names_array=[]
ratio_array=[]
def match_names(last_year,this_year):
    for row in last_year:
    x=process.extractOne(row,this_year)
    names_array.append(x[0])
    ratio_array.append(x[1])
return names_array,ratio_array


#last year company names dataset
last_year=data['Company'].dropna().values

#this year companydataset

this_year=data['Company name'].values

name_match,ratio_match=match_names(last_year,this_year)

data['this_year']=pd.Series(name_match)
data['match_rating']=pd.Series(ratio_match)

data.to_csv("test.csv")

但是，每次我执行这部分代码时，我创建的两个添加的列都没有显示在 csv 中。事实上，"test.csv" 只是和以前一样的数据框，尽管计算机显示它是最近创建的。如果有人能指出问题或以任何方式帮助我，我将不胜感激。

编辑（数据框预览）：

          Company                Company name
0                   BODYPHLO  SPORTIQUE                         NaN
1                        JOSEPH A PERRY                         NaN
2                PCH RESORT TENNIS SHOP                         NaN
3              GREYSTONE GOLF CLUB INC.                         NaN
4                 MUSGROVE COUNTRY CLUB                         NaN
5           CITY OF PELHAM RACQUET CLUB                         NaN
6                 NORTHRIVER YACHT CLUB                         NaN
7                           LAKE FOREST                         NaN
8                   TNL TENNIS PRO SHOP                         NaN
9                SOUTHERN ATHLETIC CLUB                         NaN
10           ORANGE BEACH TENNIS CENTER                         NaN

然后公司条目（去年公司数据集）结束后，"Company name"列（今年公司数据集）开始为：

4168                                NaN                LEWIS TENNIS
4169                                NaN          CHUCKS PRO SHOP AT
4170                                NaN                CHUCK KINYON
4171                                NaN   LAKE COUNTRY RACQUET CLUB
4172                                NaN   SPORTS ACADEMY & RAC CLUB

Answer 1

考虑到一列仅从另一列开始一次，您的数据帧结构很奇怪，但我们可以让它工作。让我们为您提供的 data 使用以下示例数据框：

                        Company               Company name
0           BODYPHLO  SPORTIQUE                        NaN
1                JOSEPH A PERRY                        NaN
2        PCH RESORT TENNIS SHOP                        NaN
3      GREYSTONE GOLF CLUB INC.                        NaN
4         MUSGROVE COUNTRY CLUB                        NaN
5   CITY OF PELHAM RACQUET CLUB                        NaN
6         NORTHRIVER YACHT CLUB                        NaN
7                   LAKE FOREST                        NaN
8           TNL TENNIS PRO SHOP                        NaN
9        SOUTHERN ATHLETIC CLUB                        NaN
10   ORANGE BEACH TENNIS CENTER                        NaN
11                          NaN               LEWIS TENNIS
12                          NaN         CHUCKS PRO SHOP AT
13                          NaN               CHUCK KINYON
14                          NaN  LAKE COUNTRY RACQUET CLUB
15                          NaN  SPORTS ACADEMY & RAC CLUB

然后执行匹配：

import pandas as pd
from fuzzywuzzy import process, fuzz

known_list = data['Company name'].dropna()

def find_match(x):

    match = process.extractOne(x['Company'], known_list, scorer=fuzz.partial_token_sort_ratio)
    return pd.Series([match[0], match[1]])

data[['this year','match_rating']] = data.dropna(subset=['Company']).apply(find_match, axis=1, result_type='expand')

产量：

                        Company Company name                  this year  \
0           BODYPHLO  SPORTIQUE          NaN  SPORTS ACADEMY & RAC CLUB   
1                JOSEPH A PERRY          NaN         CHUCKS PRO SHOP AT   
2        PCH RESORT TENNIS SHOP          NaN               LEWIS TENNIS   
3      GREYSTONE GOLF CLUB INC.          NaN  LAKE COUNTRY RACQUET CLUB   
4         MUSGROVE COUNTRY CLUB          NaN  LAKE COUNTRY RACQUET CLUB   
5   CITY OF PELHAM RACQUET CLUB          NaN  LAKE COUNTRY RACQUET CLUB   
6         NORTHRIVER YACHT CLUB          NaN  LAKE COUNTRY RACQUET CLUB   
7                   LAKE FOREST          NaN  LAKE COUNTRY RACQUET CLUB   
8           TNL TENNIS PRO SHOP          NaN               LEWIS TENNIS   
9        SOUTHERN ATHLETIC CLUB          NaN  SPORTS ACADEMY & RAC CLUB   
10   ORANGE BEACH TENNIS CENTER          NaN               LEWIS TENNIS   

    match_rating  
0           47.0  
1           43.0  
2           67.0  
3           43.0  
4           67.0  
5           72.0  
6           48.0  
7           64.0  
8           67.0  
9           50.0  
10          67.0

使用 Python 模糊匹配同一数据框中的两列

Fuzzy Matching Two Columns in the Same Dataframe Using Python

python

pandas

fuzzywuzzy