Update pandas dataframe values ​obtained from a function

Question

my goal is to update my 'df1' dataframe with the new 'df2' values ​​that are produced by a function. This is the simplification of a more complex script and I cannot exempt myself from using two functions.

import pandas as pd

def new_df2(i):
    d2 = {'col1': [i, i-1], 'col2': [i+1, i+2]}
    df2 = pd.DataFrame(data=d2)
    merge_df(df1, df2)

def merge_df(df1, df2):
    df1 = df1.append(df2)

d1 = {'col1': [5, 6], 'col2': [7, 8]}
df1 = pd.DataFrame(data=d1)

for i in range(1,3,1):
    new_df2(i)

my result:

print(df1)

   col1  col2
0     5     7
1     6     8

expected result:

print(df1)

   col1  col2
0     5     7
1     6     8
0     1     2
1     0     3
0     2     3
1     1     4

Answer 1

You are having a misunderstanding regarding global and local variables for functions and I would recommend you research this topic. Functions work primarily with local variables that get passed into them. So you have to follow your dataframe in and return it out as a variable. Here is the working code and I'll comment further below.

def new_df2(i, df1):
    d2 = {"col1": [i, i - 1], "col2": [i + 1, i + 2]}
    df2 = pd.DataFrame(data=d2)
    return merge_df(df1, df2)


def merge_df(df1, df2):
    return df1.append(df2)


d1 = {"col1": [5, 6], "col2": [7, 8]}
df1 = pd.DataFrame(data=d1)

for i in range(1, 3, 1):
    df1 = new_df2(i, df1)
print(df1)
   col1  col2
0     5     7
1     6     8
0     1     2
1     0     3
0     2     3
1     1     4

Starting at the bottom you have to pass in the existing dataframe you want to modify as a variable.

for i in range(1, 3, 1):
    df1 = new_df2(i, df1)

Then in new_df2 it gets accepted as a variable here:

def new_df2(i, df1):

Then you create df2 and then send both as variables to merge_df here:

return merge_df(df1, df2)

The return will return the result of the merge after it is complete.

merge_df accepts the variables here:

def merge_df(df1, df2):
    return df1.append(df2)

And returns them to new_df2 which immediately returns it to the initial call. Then we set df1 to the new value returned, and move to the next step in the loop.

Answer 2

Something like this would work:

In [232]: d1 = {'col1': [5, 6], 'col2': [7, 8]}                                                                                                                                                             

In [233]: df1 = pd.DataFrame(data=d1)                                                                                                                                                                       

In [234]: df = pd.DataFrame()                                                                                                                                                                               

In [235]: def new_df2(i): 
     ...:     d2 = {'col1': [i, i-1], 'col2': [i+1, i+2]} 
     ...:     df2 = pd.DataFrame(data=d2)  
     ...:     return df2  
     ...:                                                                                                                                                                                                   

In [236]: def merge_df(df1, df2): 
     ...:     df1 = df1.append(df2)  
     ...:     return df1  
     ...:                                                                                                                                                                                                   

In [237]: for i in range(1,3,1): 
     ...:     df2 = new_df2(i) 
     ...:     df = df.append(df2)  
     ...:                                                                                                                                                                                                   

In [238]: df1 = merge_df(df1,df)                                                                                                                                                                            

In [239]: df1                                                                                                                                                                                               
Out[239]: 
   col1  col2
0     5     7
1     6     8
0     1     2
1     0     3
0     2     3
1     1     4