DataFrame populated with the letter "n" when no data passed and dtype is specified as str

[ajcr]

This is discussed on Stack Overflow here.

When constructing a DataFrame with the dtype as str but not passing in any data (just an index), the resulting DataFrame is filled with the letter n:

In [61]: pd.DataFrame(index=range(2), columns=[0], dtype=str)
Out[61]: 
   0
0  n
1  n

This is contrary to the behaviour of Series when no data is passed in:

In [62]: pd.Series(index=range(2), dtype=str)
Out[62]: 
0    NaN
1    NaN
dtype: object

It would be logical for the DataFrame to be populated with NaN in this instance too.

[jreback]

closed by #10658