Pandas DataFrame | convert_dtypes method
Start your free 7-days trial now!
Pandas DataFrame.convert_dtypes(~) method converts the data type of the columns of the source DataFrame to a more suitable/specific one.
This method returns a new DataFrame, that is, the source DataFrame is kept intact.
Parameters
1. infer_objects | boolean | optional
Whether or not to convert object dtypes to a more suitable and specific dtype. By default, infer_objects=True.
2. convert_string | boolean | optional
Whether or not to convert object dtypes to a string dtype. By default, convert_string=True.
3. convert_integer | boolean | optional
Whether or not to convert to integer dtypes, if possible. By default, convert_integer=True.
4. convert_boolean | boolean | optional
Whether or not to convert object types to boolean dtype. By default, convert_boolean=True.
Return Value
A DataFrame with the converted dtypes.
Examples
Consider the following DataFrame:
df = pd.DataFrame({"A":["alex",np.NaN, "bob"],"B":[10,20,np.NaN], "C":[10.5,15.0,np.NaN], "D":[np.NaN,True,False]})df
A B C D0 alex 10.0 10.5 NaN1 NaN 20.0 15.0 True2 bob NaN NaN False
Our df has the following dtypes:
df.dtypes
A objectB float64C float64D objectdtype: object
Here, note the following:
we end up with
objecttypes for columnsAandDeven though they holdstringandbooleanvalues, respectively. This is because they containNaN, which ends up making the entire column typeobject.column
Bcan be represented by aintinstead of afloat.
To make the data types of df more specific, use the convert_dtypes(~) method:
df_converted = df.convert_dtypes()df_converted
A B C D0 alex 10 10.5 <NA>1 <NA> 20 15.0 True2 bob <NA> NaN False
Now, the converted data type are as follows:
df_converted.dtypes
A stringB Int64C float64D booleandtype: object
Here, note the following:
the types of columns
AandDhave been converted tostringandboolean, respectively.column
Bis nowintinstead offloat.
