How to concatenate two pandas dataframes after replacing data within one function

Question

I'm trying to write a function which takes a dataframe, duplicates it, changes the values in one of the columns and then concatenates them together. For example:

df1 looks like

column1 | column2 | column3
keyword1|    1    |   20
keyword2|    1    |   40
keyword3|    1    |   60

and I want a function which creates this:

column1 | column2 | column3
keyword1|    1    |   20
keyword2|    1    |   40
keyword3|    1    |   60
keyword1|    2    |   20
keyword2|    2    |   40
keyword3|    2    |   60

I've tried this:

def add_mobile(df):

    mobiledf = df.copy(deep=True)
    mobiledf['column2'].replace(1,2,inplace=True)

    df = pd.concat([df,mobiledf],axis=0)

however when I put df1 though this function it returns df1 unedited. If I then look at df1.shape it has not changed.

Can you try return df below the function? Because I have tried your code and it worked for me. — demirbilek, Jun 08 '18 at 13:25
The second function just returns the copied and modified `df`. Do you assign it to variable and perform the concat afterwards? — Fourier, Jun 08 '18 at 13:25
The second function is there to show that the second DF is being made with no problems. The first function does concatenate them. — jceg316, Jun 08 '18 at 13:26
You are not returning your df in first function and the second function does not concat the dataframes. — harvpan, Jun 08 '18 at 13:29
If I put a return statement in my function it works, however when I add df1 as a parameter it doesn't actually edit df1. — jceg316, Jun 08 '18 at 13:29

score 1 · Answer 1 · answered Jun 08 '18 at 13:27

1

This worked for me:

def add_mobile(df):
   mobiledf = df.copy(deep=True)
   mobiledf['column2'].replace(1,2,inplace=True)
   df = pd.concat([df,mobiledf],axis=0,ignore_index=True)
   return df

answered Jun 08 '18 at 13:27

demirbilek

331
1
11

Emmanuel-Lin · Accepted Answer · 2018-06-08T14:02:32.507

1

First function

You didn't return df so it didn't do anything:

def add_mobile(df):

    mobiledf = df.copy(deep=True)
    mobiledf['column2'].replace(1,2,inplace=True)

    df = pd.concat([df,mobiledf],axis=0)

    return df

Second function:

You didn't concatenate:

def add_mobile(df):

    mobiledf = df.copy(deep=True)
    mobiledf['column2'].replace(1,2,inplace=True)

    return pd.concat([df,mobiledf],axis=0)

Concatenate is not an in-place function

Finlay, concatenate is not an in-place function, so you should always retrieve the result:

df = add_mobile(df)

edited Jun 08 '18 at 14:02

answered Jun 08 '18 at 13:29

Emmanuel-Lin

1,848
1
16
31

This function works. But when I look at `df1.shape` it remains the same and does not have the other df concatenated. It's as if there's an `inplace=True` missing or something – jceg316 Jun 08 '18 at 13:35
1

Yes, this is because concat is not an inplace function. Just perform `df1 = add_mobile(df1)` – Emmanuel-Lin Jun 08 '18 at 13:42
Are have a look at this [question about concatenating without copy](https://stackoverflow.com/questions/7869095/concatenate-numpy-arrays-without-copying) – Emmanuel-Lin Jun 08 '18 at 13:43

How to concatenate two pandas dataframes after replacing data within one function

2 Answers2