CODEWITHSUNDEEP
pythonpandastypesrolling-computation
Bill
Bill

Reputation: 11603

Why does Pandas rolling method return a series with a different dtype to the original?

Just curious why the Pandas Series rolling window method doesn't preserve the data-type of the original series:

import numpy as np
import pandas as pd

x = pd.Series(np.ones(6), dtype='float32')
x.dtype, x.rolling(window=3).mean().dtype

Output:

(dtype('float32'), dtype('float64'))

Upvotes: 1

Views: 49

Answers (1)

tdelaney
tdelaney

Reputation: 77337

x.rolling(window=3) gives you a pandas.core.window.rolling.Rolling object. help(pandas.core.window.rolling.Rolling.mean) includes the note:

 Returns
-------
Series or DataFrame
    Return type is the same as the original object with ``np.float64`` dtype.

that's the little why. The big why it would do such a thing, I don't know. Perhaps it's a way to keep from losing precision since you can always choose to convert to float32 again.

Upvotes: 1

Related Questions