CODEWITHSUNDEEP
pythonpandasconditional-statementsnan
running man
running man

Reputation: 1467

Pandas add a column and fill values by using list comprehension if else

I want to add a column name A in df, and fill the values. If column a1 has value, fill df.A with df.a1, else fill df.A with df.a2.

tm = pd.DataFrame({'a1':['astr1',np.nan,'astr2',np.nan],'a2':[np.nan,np.nan,np.nan,'astr3']})
tm

    a1      a2
0   str1    NaN
1   NaN     NaN
2   str2    NaN
3   NaN     str2

I want this.

    a1      a2       A
0   str1    NaN      str1
1   NaN     NaN      NaN
2   str2    NaN      str2
3   NaN     str2     str2

Upvotes: 2

Views: 1388

Answers (2)

Quickbeam2k1
Quickbeam2k1

Reputation: 5437

In addition to jezraels answer, you can also use pandas assign function. For example his fillna solution maybe recast in the form

tm.assign(A = lambda x:x.a1.fillna(x.a2))

This may be advantageous in pandas pipelines

Upvotes: 3

jezrael
jezrael

Reputation: 863166

You can use numpy.where with mask created by isnull:

tm['A'] = np.where(tm.a1.isnull(), tm.a2, tm.a1)
print (tm)

      a1     a2      A
0  astr1    NaN  astr1
1    NaN    NaN    NaN
2  astr2    NaN  astr2
3    NaN  astr3  astr3

Another solutions with combine_first or fillna:

tm['A'] = tm.a1.combine_first(tm.a2)
print (tm)
      a1     a2      A
0  astr1    NaN  astr1
1    NaN    NaN    NaN
2  astr2    NaN  astr2
3    NaN  astr3  astr3

tm['A'] = tm.a1.fillna(tm.a2)
print (tm)
      a1     a2      A
0  astr1    NaN  astr1
1    NaN    NaN    NaN
2  astr2    NaN  astr2
3    NaN  astr3  astr3

And last solution with update:

tm['A'] = tm.a1
tm.A.update(tm.a2)
print (tm)
      a1     a2      A
0  astr1    NaN  astr1
1    NaN    NaN    NaN
2  astr2    NaN  astr2
3    NaN  astr3  astr3

Upvotes: 3

Related Questions