Change Data Type for one or more columns in Pandas Dataframe

Let’s see the different ways of changing Data Type for one or more columns in Pandas Dataframe.

We can pass any Python, Numpy or Pandas datatype to change all columns of a dataframe to that type, or we can pass a dictionary having column names as keys and datatype as values to change type of selected columns.

# importing pandas as pd 
import pandas as pd 
  
# sample dataframe 
df = pd.DataFrame({ 
    'A': [1, 2, 3, 4, 5], 
    'B': ['a', 'b', 'c', 'd', 'e'], 
    'C': [1.1, '1.0', '1.3', 2, 5] }) 
  
# converting all columns to string type 
df = df.astype(str) 
print(df.dtypes) 

Output:

# importing pandas as pd 
import pandas as pd 
  
# sample dataframe 
df = pd.DataFrame({ 
    'A': [1, 2, 3, 4, 5], 
    'B': ['a', 'b', 'c', 'd', 'e'], 
    'C': [1.1, '1.0', '1.3', 2, 5] }) 
  
# using dictionary to convert specific columns 
convert_dict = {'A': int, 
                'C': float
               } 
  
df = df.astype(convert_dict) 
print(df.dtypes) 

Output:

Method #2: Using DataFrame.apply()

We can pass pandas.to_numeric, pandas.to_datetime and pandas.to_timedelta as argument to apply() function to change the datatype of one or more columns to numeric, datetime and timedelta respectively.

# importing pandas as pd 
import pandas as pd 
  
# sample dataframe 
df = pd.DataFrame({ 
    'A': [1, 2, 3, '4', '5'], 
    'B': ['a', 'b', 'c', 'd', 'e'], 
    'C': [1.1, '2.1', 3.0, '4.1', '5.1'] }) 
  
# using apply method 
df[['A', 'C']] = df[['A', 'C']].apply(pd.to_numeric) 
print(df.dtypes) 

Output:

Method #3: Using DataFrame.infer_objects()
This method attempts soft-conversion by inferring data type of ‘object’-type columns. Non-object and unconvertible columns are left unchanged.

# importing pandas as pd 
import pandas as pd 
  
# sample dataframe 
df = pd.DataFrame({ 
    'A': [1, 2, 3, 4, 5], 
    'B': ['a', 'b', 'c', 'd', 'e'], 
    'C': [1.1, 2.1, 3.0, 4.1, 5.1] 
     }, dtype ='object') 
  
# converting datatypes 
df = df.infer_objects() 
print(df.dtypes) 

Output:

My Personal Notes

Recommended Posts: