自定义函数获取数据框值的位置

时间:2018-06-24 18:05:26

标签: python dataframe

这个问题或多或少与我上一个问题类似,但是我有一个DF

Index   Batch   Name    List Name
0        1      Jon     Adam
1           
2        2      Adam    Sam
3                       Chris
4        3      Voges   Jon
5           
6        4      Jon     Voges

我想在列表名称中搜索每个值的批号,即Adam,Sam,Chris,Jon和Voges。我想从中获得另一个DF

Index   Batch   Name    List Name   BatchNames
0        1      Jon     Adam        Adam(2)
1               
2        2      Adam    Sam         Sam(2)
3                       Chris       Chris(2)
4        3     Voges    Jon         Jon(1,4)
5               
6        4     Jon      Voges       Voges(3)

我想选择每个列表名称,并在名称中搜索它们具有的相应批号,即Jon存在于1 and 4中,依此类推。但是,如果“名称”中不存在“列表名称”中的名称,则应选择与之接近的相应批号,例如,“名称中不存在Sam,但与Batch 2接近”, Chris也是如此。基本上,批次之间的任何内容都属于最低的批次号。我如何为此编写自定义函数

1 个答案:

答案 0 :(得分:1)

我会做这样的事情:

import pandas as pd
import numpy as np

def custom_function(df):
    # Forward fill the Batch number
    df_Batch = df.Batch.copy()
    df.Batch.ffill(inplace=True)
    df.Batch = df.Batch.astype(int)
    # Make a new dataframe where we first get batches for the name column
    # and append batches for the list name column, there we be duplicates so we keep the first entry
    a = df.groupby('Name').Batch.apply(tuple).append(df.groupby('List Name').Batch.apply(tuple)).reset_index().groupby('index').first()
    # Create a series which concatenates the Batch number and List Name
    b = pd.Series(a.index.astype(str) + a.Batch.astype(str), index=a.index).replace(',','', regex=True).replace(' ',',',regex=True)
    # undo the forward fill (replace with original columns)
    df.Batch = df_Batch
    # join the series we just made to the dataframe
    return df.merge(b.to_frame().rename_axis('List Name'), how='left', on='List Name', suffixes=['', 'Names']).fillna('')
df = pd.DataFrame({'Batch':[1,np.nan,2,np.nan,3,np.nan,4], 'Name':['Jon',np.nan, 'Adam',np.nan, 'Voges',np.nan, 'Jon'], 'List Name':['Adam', np.nan, 'Sam', 'Chris', 'Jon', np.nan, 'Voges']})
# Out[122]: 
#    Batch   Name List Name
# 0    1.0    Jon      Adam
# 1    NaN    NaN       NaN
# 2    2.0   Adam       Sam
# 3    NaN    NaN     Chris
# 4    3.0  Voges       Jon
# 5    NaN    NaN       NaN
# 6    4.0    Jon     Voges
custom_function(df)
# Out[131]: 
#   Batch   Name List Name BatchNames
# 0     1    Jon      Adam    Adam(2)
# 1                                  
# 2     2   Adam       Sam     Sam(2)
# 3                  Chris   Chris(2)
# 4     3  Voges       Jon   Jon(1,4)
# 5                                  
# 6     4    Jon     Voges   Voges(3)