Cannot Display Styled Dataframe with NaN values

Using Streamlit
, , ,
1

I have added a colormap to a dataframe using matplotlibs colormaps.

df.style.background_gradient(axis=None, cmap='coolwarm')

This works fine for columns with only NaN values but throws an error when a column in the dataframe has both numbers and NaN values.

The error is thrown when trying to display the styled dataframe with “st.dataframe()”

ValueError: cannot convert to ‘float64’-dtype NumPy array with missing values. Specify an appropriate ‘na_value’ for this dtype.

File "C:\Work\Systems\Maturity Assessment\engineering-maturity\Maturity.py", line 90, in create_project_rollup
    st.dataframe(heat_mapped_deviation_table, use_container_width=True, hide_index=True, height=(len(latest_assessments) + 1) * 35 + 3)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\streamlit\runtime\metrics_util.py", line 410, in wrapped_func
    result = non_optional_func(*args, **kwargs)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\streamlit\elements\arrow.py", line 546, in dataframe
    marshall_styler(proto, data, default_uuid)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\streamlit\elements\lib\pandas_styler_utils.py", line 61, in marshall_styler
    styler._compute()
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\io\formats\style_render.py", line 258, in _compute
    r = func(self)(*args, **kwargs)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\io\formats\style.py", line 1719, in _apply
    result = func(data, **kwargs)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\io\formats\style.py", line 3926, in _background_gradient
    gmap = data.to_numpy(dtype=float)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\core\frame.py", line 1843, in to_numpy
    result = self._mgr.as_array(dtype=dtype, copy=copy, na_value=na_value)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\core\internals\managers.py", line 1770, in as_array
    arr = self._interleave(dtype=dtype, na_value=na_value)
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\core\internals\managers.py", line 1829, in _interleave
    arr = blk.values.to_numpy(  # type: ignore[union-attr]
File "C:\Work\Tools\Anaconda3\envs\Streamlit\lib\site-packages\pandas\core\arrays\masked.py", line 412, in to_numpy
    raise ValueError(

As you can see, Streamlit is calling to Numpy to convert the dataframe to a numpy array with “data.to_numpy(dtype=float)” which cannot seem to handle columns with both numbers and NaN values.

Any ideas on how to fix this? I can’t really replace the NaNs with 0 as 0 has a different meaning in my data.

2

They do handle NaN values, my guess is that data in your dataframe are not numeric. What are the df.dtypes of your dataframe?

3

This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.