ENH: Infer column dtype when constructing object with pd.NA

[amanlai]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

When a list that contains integers and pd.NAs is cast in a pandas object, the column becomes dtype=object. An example illustrating the issue:

Feature Description

pd.Series([pd.NA, 1]).dtype            # object

Alternative Solutions

Is it possible to cast it into an Extension Dtype such as 'Int64' or at least a numeric dtype such as float64 or Float64?

Additional Context

No response

[asishm]

You can use pd.Series.convert_dtypes to achieve this.

In [39]: pd.Series([pd.NA, 1]).convert_dtypes().dtype
Out[39]: Int64Dtype()

That said, None does get converted to nan in pd.Series([None, 1]) and gets assigned a float dtype. pd.NA having similar behavior would make sense.

[mroeschke]

#58243 is discussing the path for nullable types being returned by default.

Having one sentinel changing the returned dtype leads to value dependent behavior which pandas is trying to avoid, so I think this change would be better suited for the migration in #58243