Pandas DataFrame: Change column values?

user_14004586 0
Viewed 1259

I have a df where two column values have True, False or NaN

df

            a         b           c

    0      a         True        NaN
    1      b         False       True
    2      c         NaN         False
    3      d         NaN         NaN
    4      e         False       NaN
    5      f         True        False

I'm trying to convert the values in columns b and c.

If True comes in one column, another will be False If NaN present change that False.

Similarly for False. If NaN is present in both columns. Change both values to False

Resultant df:

            a         b           c

    0      a         True        False
    1      b         False       True
    2      c         True        False
    3      d         False       False
    4      e         False       True
    5      f         True        False
pandas dataframe python
5 Answers

Let's try two-step filling:

s = df[['b','c']]

# both are `NaN`, fill with `False`
df.loc[s.isna().all(1), ['b','c']] = False

# inverse the sum 
sums = (1 - s.sum(1)).astype(bool)

# fill the remaining `NaN` with the inverse sum
df[['b','c']] = s.apply(lambda x: x.fillna(sums))

Output:

   a      b      c
0  a   True  False
1  b  False   True
2  c   True  False
3  d  False  False
4  e  False   True
5  f   True  False
user_42384080

There are three inbuilt python methods to change column values based on condition. You can refer to the below link for same: https://pythonexamples.org/pandas-dataframe-replace-values-in-column-based-on-condition/

My problem statement was for the column ACTIONCODE in data frame (df2) if column value is A/AMEND/correction then change it to MODIFY

df2.loc[((df2.ACTIONCODE == 'A') | (df2.ACTIONCODE == 'AMEND') | (df2.ACTIONCODE == 'correction')), 'ACTIONCODE'] = 'MODIFY'

NOTE: The or and and python statements require truth-values. For pandas these are considered ambiguous so you should use “bitwise” | (or) or & (and) operations

user_140213500

This uses numpy where to check each column for the conditions and replace with the opposite value :

cond1 = df.b.notna() & df.c.isna()
cond2 = df.b.isna() & df.c.notna()

df.assign(
    c=lambda x: np.where(cond1, df.b.sub(1).ne(0), df.c),
    b=lambda x: np.where(cond2, df.c.sub(1).ne(0), df.b),
).mask(lambda x: x.isna(), False)


    a   b        c
0   a   True    False
1   b   False   True
2   c   True    False
3   d   False   False
4   e   False   True
5   f   True    False

Alternatively, you could use .loc and assign the values :

cond3 = df.b.isna() & df.c.isna()

df.loc[cond1, "c"] = df.loc[cond1, "b"].sub(1).ne(0)
df.loc[cond2, "b"] = df.loc[cond2, "c"].sub(1).ne(0)
df.loc[cond3, ["b", "c"]] = False
user_71757130

create a user-defined function swap:

import numpy as np


def swap(x):
  if(x['b']==False and x['c']==np.NaN):
    return True
  if(x['b']==True and x['c']==np.NaN):  
    return False
  else:
    return c  
 def swap2(x):
      if(x['c']==False and x['b']==np.NaN):
        return True
      if(x['c']==True and x['b']==np.NaN):  
        return False
      else:
        return b

Then use apply function:

df['c']=df.apply(swap, axis=1)

df['b']=df.apply(swap2, axis=1)

then fill NAs with false

df.fillna(False)
user_59067750

Create filters for conditions

na_false = (df['b'].isna()) & (df['c'] == False)

na_true = (df['b'].isna()) & (df['c'] == True)

true_na = (df['b'] == True) & (df['c'].isna())

false_na = (df['b'] == False) & (df['c'].isna())

na_na = (df['b'].isna()) & (df['c'].isna())

Set column values using filters

df.loc[na_false,'b'] = True

df.loc[na_true,'b'] = False

df.loc[true_na, 'c'] = False

df.loc[false_na, 'c'] = True

df.loc[na_na, ['b', 'c']] = False
user_129690010
Related