1

Hi I have an excel data with multiple columns and i need to fined specific word and return it in new column the table look like this:

ID   col0  col1  col2  col3  col4  col5
1    jack  a/h   t/m   w/n   y/h    56
2    sam   z/n   b/w   null  null   93
3    john  b/i   y/d   p/d   null   33

I want to look for 'b' in columns col1, col2, col3, and col4 and create a new column called "b" where the value the cell value with be is returned

the result would look like this

ID   col0  col1  col2  col3  col4  col5  b
1    jack  a/h   t/m   w/n   y/h    56   -
2    sam   z/n   b/w   null  null   93   b/w
3    john  b/i   y/d   p/d   null   33   b/i

and I need an efficient way to do it I tried to use where like this

df1 = df[['col1', 'col2', 'col3', 'col4']]

df1['b']==[x for x in df1.values[0] if any(b for b in lst if b in str(x))]

I got this from this answer https://stackoverflow.com/a/50250103/3105140

yet it is not working for me snice I have null value and rows where the condition do not work

SMO
  • 149
  • 8

3 Answers3

3

Here is a way using stack and str.contains with df.where:

cols = ['col1', 'col2', 'col3', 'col4']
df['b'] = (df[cols].where(df[cols].stack().str.contains('b')
         .unstack(fill_value=False)).ffill(1).iloc[:,-1])

print(df)

   ID  col0 col1 col2 col3 col4  col5    b
0   1  jack  a/h  t/m  w/n  y/h    56  NaN
1   2   sam  z/n  b/w  NaN  NaN    93  b/w
2   3  john  b/i  y/d  p/d  NaN    33  b/i
anky
  • 74,114
  • 11
  • 41
  • 70
3

I would use DataFrame.stack with callable:

cols = ['col1', 'col2', 'col3', 'col4']
df['b']=(df[cols].stack()
                 .loc[lambda x: x.str.contains('b')]
                 .reset_index(level=1,drop=1)
                #.fillna('-') #for the expected output
        )

Output

   ID  col0 col1 col2 col3 col4  col5    b
0   1  jack  a/h  t/m  w/n  y/h    56  NaN
1   2   sam  z/n  b/w  NaN  NaN    93  b/w
2   3  john  b/i  y/d  p/d  NaN    33  b/i
ansev
  • 30,322
  • 5
  • 17
  • 31
0

In a bid to avoid selecting the columns, I used melt:

M = (df.copy()
     .melt(id_vars='ID')
     .loc[lambda x:x['value'].astype('str').str.contains('b')]
     .drop('variable',axis=1))

pd.merge(df,M,how='left',on='ID').rename({'value':'b'},axis=1)

    D   col0    col1    col2    col3    col4    col5     b
0   1   jack    a/h     t/m     w/n     y/h      56     NaN
1   2   sam     z/n     b/w     NaN     NaN      93     b/w
2   3   john    b/i     y/d     p/d     NaN      33     b/i
sammywemmy
  • 27,093
  • 4
  • 17
  • 31