Nested If-else pandas dataframe

Question

I have an input dataframe df, where I want to transform column A...

A   B
    asd
0   dsf
1   ewr
2   dfds
3   sdf

...into my expected output df1:

        Aa  Ab  Ac              Bb
                                asd
        0   0   0               dsf
        1   0   0               ewr
        1   1   0               dfds
        1   1   1               sdf

Code:

if df['A'] == '' :
    df1['Aa'] = ''
elif df['A'] == 1 :
    df1['Aa'] == 1
elif df['A'] == 0 :
    df1['Aa'] == 0
else:
    df1['Aa'] == 1

Error:

if df['A'] == '' :                                                                                                                                                       
  File "C:\Python\Python38\lib\site-packages\pandas\core\generic.py", line 1478, in __nonzero__                                                                                              
    raise ValueError(                                                                                                                                                                        
ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

Similar formula in excel

 B2 =IF(A2=1,1,IF(A2="","",IF(A2=0,0,1)))

Similarly;

 C2 = IF(A2=2,1,IF(A2="","",IF(A2=0,0,1)))

Trying to achieve something similar with python. Please Help!!

Logic :

if value in df['A'] == ''; new column 'Aa,Ab,Ac' in df1 == ''
if value in df['A'] == 1; new column 'Aa' in df1 == 1, and rest (Ab,Ac) == 0
if value in df['A'] == 2; new column 'Aa, Ab' in df1== 1, and rest (Ac) == 0
if value in df['A'] == 3; new column 'Aa,Ab,Ac' in df1 == 1

So, the flow can be something like:

df1:

A   Aa  Ab  Ac              Bb
                            asd
0   0   0   0               dsf
1   1   0   0               ewr
2   1   1   0               dfds
3   1   1   1               sdf

Later just remove column A. So Final df1

        Aa  Ab  Ac              Bb
                                asd
        0   0   0               dsf
        1   0   0               ewr
        1   1   0               dfds
        1   1   1               sdf

Do the values in Aa, Ab, Ac actually depend on the original values in A beyond carrying the empty string value across the whole row? Or do you just want 1s and 0s arranged in this triangular pattern? — G. Anderson
– G. Anderson, Commented Oct 6, 2020 at 19:48
Have a look into the numpy.where() function. Here are the docs. It adds excellent 'if/then/else' logic to DataFrames. — s3dev
– s3dev, Commented Oct 6, 2020 at 20:30
Your input column df.A is numeric, so trying to do a string compare df['A'] == '' will also fail for that reason, as well as the issue you get with the output being a series not just a scalar. — smci
– smci, Commented Oct 6, 2020 at 21:42
Essentially you want to convert the 2-bit binary value in df.A into binary-encoded columns [Aa,Ab,Ac]. Although they're flipped left-to-right; normally encoding would start from the LSB, i.e. Ac. — smci
– smci, Commented Oct 6, 2020 at 21:45
@G.Anderson They do depend on A, this is just an example. In my dataset I have multiple 1s,2s and so on... based on another logic so A could be like 0 0 0 0 1 1 3 2 2... . I have provided the logic that I need for calculation of Aa Ab Ac{random column names}, just can't determine how to code this — RoshanShah22
– RoshanShah22, Commented Oct 7, 2020 at 5:56

RoshanShah22 · Accepted Answer · 2020-10-07 06:21:35Z

3

Found a way to do this, just put this inside a function, and you could use it for the other nested loops too

   conditions = [
        (df['A'] == 0),
        (df['A'] == 1),
        (df['A'] > 1)]
    choices = [0, 1, 1]
    df1['Aa'] = np.select(conditions, choices, default='null')

answered Oct 7, 2020 at 6:21

RoshanShah22

4201 gold badge5 silver badges17 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Pei Li · Accepted Answer · 2020-10-06 20:33:48Z

1

IIUC, try:

import pandas as pd    
df = pd.DataFrame({'a':["",0,1,2,3]})
df['new_a'] = pd.to_numeric(df['a'], errors='coerce')
df['Aa'] = df['new_a'].apply(lambda x: np.nan if math.isnan(x) else (0 if x==0 else (OTHER CONDITION)))

you can use the similar logic to create your 'Ab', 'Ac', and 'Bb'

answered Oct 6, 2020 at 20:33

Pei Li

3203 silver badges17 bronze badges

4 Comments

El_1988 Over a year ago

Oddly I am facing the same issue today, I have many nested loops so the solution proposed wont work....am not sure why this is happening

RoshanShah22 Over a year ago

Similar situation as that of @El_1988, not sure this solution works for me

El_1988 Over a year ago

@RoshanShah22 I posted my question here: stackoverflow.com/questions/64237931/… , if a solution shows up will share here

RoshanShah22 Over a year ago

Hey @El_1988, I tried something and it seems to work. Maybe give it a shot

Collectives™ on Stack Overflow

Nested If-else pandas dataframe

2 Answers 2

Comments

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related