DOC: DataFrame.update() fails to change values in calling dataframe if new value is NaN

Code Sample, a copy-pastable example if possible

# Define dataframe
tempDF1 = pd.DataFrame({'c1':[1,2,3,4,5],
                       'c2':[True,True,False,False,True]})

# Select part of original dataframe
tempDF2 = tempDF1.loc[tempDF1['c2']==True,:].copy()

print('Original dataframe')
print(tempDF1)
print('\nSelection of dataframe')
print(tempDF2)

# Make some changes to selection
tempDF2.loc[:,'c1'] = tempDF2.loc[:,'c1'] + 100
tempDF2.iloc[2,0] = np.nan

# Update original dataframe with new values
tempDF1.update(tempDF2,overwrite=True)

print('\nSelection of dataframe - with changes')
print(tempDF2)
print('\nUpdated original dataframe')
print(tempDF1)

Problem description

DataFrame.update() function fails to update a dataframe with new NaN values. However, non-NaN values are updated to original dataframe with no issues (except the dtype of the dataframe is altered in the update process, namely int64 changed to float64).

Expected Output

Expected output would be

      c1     c2
0  101   True
1  102   True
2    3  False
3    4  False
4  NaN   True

However, actual output is:

      c1     c2
0  101.0   True
1  102.0   True
2    3.0  False
3    4.0  False
4    5.0   True

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None python: 3.4.6.final.0 python-bits: 64 OS: Darwin OS-release: 15.6.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_GB.UTF-8 LOCALE: en_GB.UTF-8

pandas: 0.20.2 pytest: None pip: 9.0.1 setuptools: 34.3.3 Cython: None numpy: 1.13.0 scipy: None xarray: None IPython: 5.3.0 sphinx: None patsy: None dateutil: 2.6.0 pytz: 2017.2 blosc: None bottleneck: None tables: None numexpr: None feather: None matplotlib: None openpyxl: 2.4.7 xlrd: None xlwt: None xlsxwriter: None lxml: None bs4: None html5lib: 0.999999999 sqlalchemy: None pymysql: 0.7.11.None psycopg2: None jinja2: 2.9.6 s3fs: None pandas_gbq: None pandas_datareader: None

[maxwellflitton]

Is there a parameter to track NaN values and change old DF values to NaN in the update function?

[joh-mue]

Yes, what if I want to force NaN values in the update function?