Creating new column from existing columns

Question

I have a data-frame

ID   P_1   P_2
1    NaN   NaN
2    124   342
3    NaN   234
4    123   NaN
5    2345  500

I want to make a new column titled P_3 such that:

ID   P_1   P_2  P_3
1    NaN   NaN   NaN
2    124   342   342
3    NaN   234   234
4    123   NaN   123
5    2345  500  500

My conditions are:

if P_1 = Nan , then P_3 == P_2
if P_1 != Nan and P_2 != Nan, then  P_3 == P_2
if P_2 = Nan , then P_3 == P_1

I have applied the following codes:

conditions = [
    (df['P_1'] == float('NaN')),
    (df['P_1'] != float('NaN')) & (df['P_2'] != float('NaN')),
    (df['P_1'] != float('NaN')) & (df['P_2'] == float('NaN'))
    ]

values = [df['P_2'], df['P_2'], df['P_1']]

df['P_3'] = np.select(conditions, values)

But it gives me the following error:

Length of values does not match length of index

Corralien · Accepted Answer · 2021-03-21 06:01:44Z

2

In summary, your unique condition is:

P_3 = P_2 if P_2 != NaN else P_1

combine_first: update null elements with value in the same location in other (ref: Pandas doc.)

>>> df["P_2"].combine_first(df["P_1"])
ID
1      NaN
2    342.0
3    234.0
4    123.0
5    500.0

edited Mar 21, 2021 at 6:01

answered Mar 21, 2021 at 5:49

Corralien

121k8 gold badges43 silver badges67 bronze badges

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

Comments

inspectorG4dget · Accepted Answer · 2021-03-21 06:10:23Z

Another approach:

In [93]: df                                                                                                                                                                                                                                                                   
Out[93]: 
       p1     p2
0     NaN    NaN
1   124.0  342.0
2     NaN  234.0
3   123.0    NaN
4  2345.0  500.0

In [94]: df['p3'] = df.p2                                                                                                                                                                                                                                                     

In [95]: df                                                                                                                                                                                                                                                                   
Out[95]: 
       p1     p2     p3
0     NaN    NaN    NaN
1   124.0  342.0  342.0
2     NaN  234.0  234.0
3   123.0    NaN    NaN
4  2345.0  500.0  500.0

In [96]: df.loc[df.p3.isna(), 'p3'] = df[df.p3.isna()]['p1']                                                                                                                                                                                                                  

In [97]: df                                                                                                                                                                                                                                                                   
Out[97]: 
       p1     p2     p3
0     NaN    NaN    NaN
1   124.0  342.0  342.0
2     NaN  234.0  234.0
3   123.0    NaN  123.0
4  2345.0  500.0  500.0

A more simpler version of your answer would be: df["P_3"] = df["P_2"].fillna(df["P_1"])

Collectives™ on Stack Overflow

Creating new column from existing columns

2 Answers 2

Comments

1 Comment

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Related