CODEWITHSUNDEEP
pythonpandasdataframe
Jake B.
Jake B.

Reputation: 465

convert list of strings to panda dataframe with types

I have a list of a list of numbers where two rows are strings, e.g.

A = [[1,'5.4','2'],[2,'6','3']]

How do I convert this to a pandas dataframe, such that the 1st and 3nd columns are integers and the 2nd column is a float

by

pd.DataFrame(A,dtype=float)

it converts all to floats.

Upvotes: 1

Views: 875

Answers (3)

Riccardo Bucco
Riccardo Bucco

Reputation: 15364

Here is a possible solution:

pd.DataFrame(A).astype({0: int, 1: float, 2: int})

If you don't want to convert everything to a string and then change the type you could do something like this (everything is loaded as a float and then later you change the dtype of a few columns to int):

pd.DataFrame(A, dtype=float).convert_dtypes()

or

pd.DataFrame(A, dtype=float).astype({0: int, 2: int})

Upvotes: 1

Corralien
Corralien

Reputation: 120409

Or use pd.to_numeric:

df = pd.DataFrame(A).apply(pd.to_numeric)

>>> df
   0    1  2
0  1  5.4  2
1  2  6.0  3

>>> df.dtypes
0      int64
1    float64
2      int64
dtype: object

If you have non numeric columns, use:

pd.DataFrame(A).apply(pd.to_numeric, errors='ignore')

Upvotes: 1

U13-Forward
U13-Forward

Reputation: 71570

You could use applymap with pd.eval:

>>> df = pd.DataFrame(A).applymap(pd.eval)
>>> df
   0    1  2
0  1  5.4  2
1  2  6.0  3
>>> df.dtypes
0      int64
1    float64
2      int64
dtype: object
>>>

Upvotes: 1

Related Questions