
        
        
            
                
                
                    df = pd.DataFrame({"A":[3,4],"B":[5,6]}, index=["a","b"])
df
                
            
                A   B
a   3   5
b   4   6

The fastest and simplest way of creating a new column is to use simple column-arithmetics:


        
        
            
                
                
                    df["C"] = df["A"] + df["B"]
df
                
            
                A   B   C
a   3   5   8
b   4   6   10

For slightly more complicated operations, use the DataFrame's native methods:


        
        
            
                
                
                    df["C"] = df.max(axis=1)
df
                
            
                A   B   C
a   3   5   5
b   4   6   6

Note the following:

we are populating the new column C with the maximum of each row (axis=1).
the return type of df.max(axis=1) is Series.

Creating binary column values

Consider the following Pandas DataFrame:


        
        
            
                
                
                    df = pd.DataFrame({'name':['Alex','Bob','Cathy'],'age':[20,30,40]})
df.head()
                
            
              name    age
0  Alex   20
1  Bob    30
2  Cathy  40

To create a new column of binary values that are based on the age column, use NumPy's where(~) method:


        
        
            
                
                
                    df['status'] = np.where(df['age'] < 25, 'JUNIOR', 'SENIOR')
df.head()
                
            
               name   age  status
0  Alex   20   JUNIOR
1  Bob    30   SENIOR
2  Cathy  40   SENIOR

Here, the first argument of the where(~) method is a boolean mask. If the boolean value is True, then resulting value will be 'JUNIOR', otherwise the value will be 'SENIOR'.

Creating column with multiple values

Once again, consider the following Pandas DataFrame:


        
        
            
                
                
                    df = pd.DataFrame({'name':['Alex','Bob','Cathy'],'age':[20,30,40]})
df.head()
                
            
              name    age
0  Alex   20
1  Bob    30
2  Cathy  40

To create a new column with multiple values based on the age column, use the apply(~) function:


        
        
            
                
                
                    def my_func(row):
    if row['age'] < 25:
        val = 'JUNIOR'
    elif row['age']  < 35:
        val = 'MID-LEVEL'
    else:
        val = 'SENIOR'
    return val

df['status'] = df.apply(my_func, axis=1)
df.head()
                
            
               name   age  status
0  Alex   20   JUNIOR
1  Bob    30   MID-LEVEL
2  Cathy  40   SENIOR

Here, the apply(~) function is iteratively called for each row, and takes in as argument a Series representing a row.

Creating column via mapping

Consider the same Pandas DataFrame as before:


        
        
            
                
                
                    df = pd.DataFrame({'name':['Alex','Bob','Cathy'],'age':[20,30,40]})
df.head()
                
            
              name    age
0  Alex   20
1  Bob    30
2  Cathy  40

To create a new column that is based on some mapping of an existing column:


        
        
            
                
                
                    mapping = {
    'Alex': 'ALEX',
    'Bob': 'BOB',
    'Cathy': 'CATHY'
}
df['upper_name'] = df['name'].replace(mapping)
df.head()
                
            
               name   age  upper_name
0  Alex   20   ALEX
1  Bob    30   BOB
2  Cathy  40   CATHY

Creating column using the assign method

Consider the following Pandas DataFrame:


        
        
            
                
                
                    df = pd.DataFrame({"A":[3,4],"B":[5,6]}, index=["a","b"])
df
                
            
               A  B
a  3  5
b  4  6

We could also use the DataFrame's assign(~) method, which takes in as argument a function with the DataFrame as the input and returns the new column values:


        
        
            
                
                
                    def foo(df):
    if df["A"].sum() > df["B"].sum():
        return [-1,-1]
    else:
        return [0,0]

df.assign(C=foo)
                
            
               A  B  C
0  3  5  0
1  4  6  0

Note the following:

if the sum of column A is larger than that of column B, then [-1,-1] will be used as the new column, otherwise [0,0] will be used.
the keyword argument (C) became the new column label.