Assign data type for each column in pandas DataFrame - Python

Question

I want to change the data type that could be entered in an empty Pandas data frame. So I've tried the following approach but it doesn't work. can someone please give me a hint how to solve this issue?

columns=['A', 'B', 'C', 'D', 'E','F']


df = pd.DataFrame(columns=columns)

>>> df
Empty DataFrame
Columns: [A, B, C, D, E, F]
Index: []

df[[0]]=df[[0]].astype(str)
df[[1]]=df[[1]].astype(str)
df[[2]]=df[[2]].astype(int)
df.iloc[:,3:6]=df.iloc[:,3:6].astype(float)

In your minimal sample, you specify some list of columns and then the resultant df doesn't show these columns, also why does this matter for an empty df? The dtype will be changed once you assign data to those columns — EdChum
– EdChum, Commented Aug 31, 2017 at 10:44
I think only is possible set same dtype in all columns by constructor like df = pd.DataFrame(columns=columns, dtype=str) — jezrael
– jezrael, Commented Aug 31, 2017 at 10:46
yes, it will work. The documentation describes it so df = pd.DataFrame(columns=columns, dtype=str) will give df.dtypes as objects — jkhadka
– jkhadka, Commented Aug 31, 2017 at 10:50

piRSquared · Accepted Answer · 2017-08-31 13:10:39Z

6

You want to construct a series or dictionary with your desired types then use astype

columns = list('ABCDEF')
df = pd.DataFrame(columns=columns)

dtypes = {k: str for k in columns[:2]}
dtypes.update({columns[2]: int})
dtypes.update({k: float for k in columns[3:]})

df = df.astype(dtypes)

df.dtypes

A     object
B     object
C      int64
D    float64
E    float64
F    float64
dtype: object

answered Aug 31, 2017 at 13:10

piRSquared

296k68 gold badges509 silver badges654 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

StingyJack Over a year ago

If you have any float NaN, using astype(str) will convert them the values to "nan"

ShreyasG · Accepted Answer · 2017-08-31 10:52:39Z

0

By default, all your columns will be of 'object' type so you might not need to force columns to be of type string as such. For the other column types, something like this might work?

df[['C']] = df[['C']].apply(pd.to_numeric)

answered Aug 31, 2017 at 10:52

ShreyasG

8065 silver badges11 bronze badges

Comments

Elletlar · Accepted Answer · 2019-02-10 22:58:46Z

0

I have also faced this problem initially but I have found a solution:

Convert the data frame column to a list data structure in Python.
Then convert the list to a series after import numpy package.
Using the astype() function convert to the desired data type.

Code:

list = list(data['unknown'])
series = pd.Series(list)
seriesOfTypeBool = g.astype(np.bool)
data['unknown'] = seriesOfTypeBool` <br>

And Simplified Version:

data['Action'] = pd.Series(list(data['Action'])).astype(np.bool)

edited Feb 10, 2019 at 22:58

Elletlar

3,2447 gold badges35 silver badges40 bronze badges

answered Feb 10, 2019 at 21:40

Sureya Pragaash

316 bronze badges

Collectives™ on Stack Overflow

Assign data type for each column in pandas DataFrame - Python

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related