how to use pandas convert_dtypes()?-CodePudding

I see quite a number of questions regarding assigning dtype, but most of them are outdated and recommending manual assignment.

Looks new method df.convert_dtypes() are available, but somehow it does not work for my case.

When I load csv files, all columns dytpe are object, and even after doing convert_dytpes(), dytpes are still object.

I would really appreciate if you can help.

df = pd.read_csv('file.csv', header=5)
df = df.convert_dtypes()

CodePudding user response：

Could you provide some data? It seems to work fine to me:

import pandas as pd
import numpy as np

print('pandas version:', pd.__version__)
print('numpy version :', np.__version__)

df = pd.DataFrame(
    {'string': pd.Series(['a', 'b', 'c'], dtype=np.dtype('O')), 
     'bool': pd.Series([True, False, True], dtype=np.dtype('O')),
     'int': pd.Series([0, 1, 2], dtype=np.dtype('O')),
     'float': pd.Series([0.0, 1.1, 2.2], dtype=np.dtype('O')),
    }
)

print(df)
print(df.dtypes)

df_new = df.convert_dtypes()

print(df_new.dtypes)

Output:

pandas version: 1.3.2
numpy version : 1.20.3

  string   bool int float
0      a   True   0   0.0
1      b  False   1   1.1
2      c   True   2   2.2

string    object
bool      object
int       object
float     object
dtype: object

string     string
bool      boolean
int         Int64
float     Float64
dtype: object

CodePudding user response：

Try this :

import pandas as pd

df = pd.DataFrame({'A': ['a', 'b', 'c'], 'B': ['d', 'e', 'f']})
print("--------DataType of DataFrame---------")
print(df.dtypes)
print("--------DataType of DataFrame after converting---------")
df1=df.convert_dtypes()
print(df1.dtypes)