转换慢速熊猫迭代到应用

Question

I have the following dataframe:

   VALUE      COUNT  RECL_2007  RECL_2008  RECL_2009  A_A  A_B  A_C  B_A  B_B  \
0    189   149.5872        503        503        500    0    0    0    0    0   
1    209  1939.6160        503        503        503    0    0    0    0    0   
2    499   617.4784        503        500        503    0    0    0    0    0   
3    585    73.0688        503        503        503    0    0    0    0    0   
4    611   133.9072        503        500        503    0    0    0    0    0   
5    645   278.7904        503        503        503    0    0    0    0    0   
6    659   138.2976        500        503        503    0    0    0    0    0   
7    719   769.5744        503        503        502    0    0    0    0    0   

   B_C  C_A  C_B  C_C  
0    0    0    0    0  
1    0    0    0    0  
2    0    0    0    0  
3    0    0    0    0  
4    0    0    0    0  
5    0    0    0    0  
6    0    0    0    0  
7    0    0    0    0 

Also, the values in columns:RECL_2007, RECL_2008 and RECL_2009 correspond to the variables A, B and C as follows:

 A = 500, B=502, C=503.

I want to fill the columns A_A...C_C using the values in the COUNT column such that the RECL_2007 value gives the first part of the column name and RECL_2009 gives the second part of the column name.

Ie, if RECL_2007 == 503 and RECL_2009 == 500, then the column is C_A and its value should be updated to whatever is in the COUNT column of that row.

Currently I am iterating through the pandas dataframe using iterrows:

for index, row in df.iterrows():    
   init = OPP_LU[row[name_init]] # Get first part of column name
   finl = OPP_LU[row[name_finl]] # Get second part of column name       
   col_name = init+'_'+finl
   df.loc[index,col_name] = row['COUNT']

This is slow, but I am not sure how to translate it into something using apply. Any hints?

Answer 1

There are two ways to do that.

• You can use apply function, but you need to do extra work here(just to simplify the work).

A dictionary to help you build naming

d={'500':'A','502':'B','503':'C'}

Function for naming

name= lambda x: "{0}_{1}".format(d[str(int(x['RECL_2007']))],d[str(int(x['RECL_2009']))])

Then, go through the items and copy count item where the name is similar.

   df["C_A"] = df.apply(lambda x: x['COUNT'] if name(x)=='C_A' else 0, axis=1)

The other solution, which is simpler is to filter the data you have, then copy count item

df.loc[(df['RECL_2007']==503) & (df['RECL_2009']==503), 'C_C']= df['COUNT']

The code would look like, this is just a quick example, you need to work on the other scenarios.

data= """VALUE,COUNT,RECL_2007,RECL_2008,RECL_2009\n189,149.5872,503,503,500\n209,939.6160,503,503,503\n499,617.4784,503,500,503\n585,73.0688,503,503,503\n611,133.9072,503,500,503\n645,278.7904,503,503,503\n659,138.2976,500,503,503\n719,769.5744,503,503,502"""

import pandas as pd

from io import StringIO

df= pd.read_csv(StringIO(data.decode('UTF-8')),sep=',' )

#First approach:    
d={'500':'A','502':'B','503':'C'}
name= lambda x: "{0}_{1}".format(d[str(int(x['RECL_2007']))],d[str(int(x['RECL_2009']))])
df['C_C']=[0]*len(df.VALUE)

df["C_A"] = df.apply(lambda x: x['COUNT'] if name(x)=='C_A' else 0, axis=1)

#Second approach:     
df.loc[(df['RECL_2007']==503) & (df['RECL_2009']==503), 'C_C']= df['COUNT']

print df

Output:

   VALUE     COUNT  RECL_2007  RECL_2008  RECL_2009       C_C       C_A
0    189  149.5872        503        503        500    0.0000  149.5872
1    209  939.6160        503        503        503  939.6160    0.0000
2    499  617.4784        503        500        503  617.4784    0.0000
3    585   73.0688        503        503        503   73.0688    0.0000
4    611  133.9072        503        500        503  133.9072    0.0000
5    645  278.7904        503        503        503  278.7904    0.0000
6    659  138.2976        500        503        503    0.0000    0.0000
7    719  769.5744        503        503        502    0.0000    0.0000