[Python] pandas test if the datatype of an input series supports nan values

I had something like the following code using pandas 1.x that new generates a warning in pandas 2:

import pandas as pd
import numpy as np

df1 = pd.DataFrame({"i":[1,2,3,4,5], "a":[2,4,6,8,10]})
df2= pd.DataFrame({"i":[2,4], "a":[3,6],"b":[4,8]})

df1["b"] = np.nan
df1.loc[df2.index,:] = df2



which throws a warning: FutureWarning: Setting an item of incompatible dtype is deprecated and will raise an error in a future version of pandas.

normally I can do this to fix:

df1['b'] = df1['b'].astype(df2.b.dtype)


but in my case I wont know the datatype of column b ahead of time, so I wont know if the datatype supports nan. Is there a nice way of settings df1's dtype as either df2's datatype or using something like convert_dtypes() only as a fallback?

I know I can manually compare the column with a copy of itself with nan to see if the dtype changes but it seems computationally expensive

Continue reading...