Matplotlib/Seaborn: Where am I making an elementwise comparison? ValueError: The truth value of an array with more than one element is ambiguous

Question

toyData3 is a Pandas dataframe. sns is Seaborn module. I am able to plot the data but I get the following error when I run the code (minimal workin example) in Jupyter/Ipython console, that I don't understand. Where did I make an elementwise comparison? What is going on, and is it worth my concern?

In [3]: import pandas as pd                                                                                                                                                                                        

In [4]: import seaborn as sns                                                                                                                                                                                      

In [5]: data=[(0.  ,  0.        , 0.        ),(0.01,  0.10283333, 0.29774878), (0.02, -0.0395    , 0.16226045), (0.03,  0.06841667, 0.3111277 ), (0.04, -0.03508333, 0.19214552)]                                  

In [6]: toyData3 = pd.DataFrame.from_records(data, columns=['time', 'mean', 'SD'])                                                                                                                                 

In [7]: print(toyData3)                                                                                                                                                                                            
   time      mean        SD
0  0.00  0.000000  0.000000
1  0.01  0.102833  0.297749
2  0.02 -0.039500  0.162260
3  0.03  0.068417  0.311128
4  0.04 -0.035083  0.192146

In [8]: sns.relplot(data=toyData3, x=toyData3.index, y=toyData3.loc[:, 'mean'],  kind='line')                                                                                                                      
/home/user/anaconda3/envs/ilahi/lib/python3.8/site-packages/pandas/core/indexes/base.py:122: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
  result = op(self.values, np.asarray(other))
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-8-35f5a3329860> in <module>
----> 1 sns.relplot(data=toyData3, x=toyData3.index, y=toyData3.loc[:, 'mean'],  kind='line')

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/seaborn/relational.py in relplot(x, y, hue, size, style, data, row, col, col_wrap, row_order, col_order, palette, hue_order, hue_norm, sizes, size_order, size_norm, markers, dashes, style_order, legend, kind, height, aspect, facet_kws, **kwargs)                                                                                                                              
   1693 
   1694     # Draw the plot
-> 1695     g.map_dataframe(func, x, y,
   1696                     hue=hue, size=size, style=style,
   1697                     **plot_kws)

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/seaborn/axisgrid.py in map_dataframe(self, func, *args, **kwargs)
    829 
    830         # Finalize the annotations and layout
--> 831         self._finalize_grid(args[:2])
    832 
    833         return self

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/seaborn/axisgrid.py in _finalize_grid(self, axlabels)
    852     def _finalize_grid(self, axlabels):
    853         """Finalize the annotations and layout."""
--> 854         self.set_axis_labels(*axlabels)
    855         self.set_titles()
    856         self.fig.tight_layout()

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/seaborn/axisgrid.py in set_axis_labels(self, x_var, y_var)
    878         if x_var is not None:
    879             self._x_var = x_var
--> 880             self.set_xlabels(x_var)
    881         if y_var is not None:
    882             self._y_var = y_var

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/seaborn/axisgrid.py in set_xlabels(self, label, **kwargs)
    889             label = self._x_var
    890         for ax in self._bottom_axes:
--> 891             ax.set_xlabel(label, **kwargs)
    892         return self
    893 

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/matplotlib/axes/_axes.py in set_xlabel(self, xlabel, fontdict, labelpad, loc, **kwargs)
    245         elif loc == 'right':
    246             kwargs.update(x=1, horizontalalignment='right')
--> 247         return self.xaxis.set_label_text(xlabel, fontdict, **kwargs)
    248 
    249     def get_ylabel(self):

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/matplotlib/axis.py in set_label_text(self, label, fontdict, **kwargs)
   1561         """
   1562         self.isDefault_label = False
-> 1563         self.label.set_text(label)
   1564         if fontdict is not None:
   1565             self.label.update(fontdict)

~/anaconda3/envs/ilahi/lib/python3.8/site-packages/matplotlib/text.py in set_text(self, s)
   1163         if s is None:
   1164             s = ''
-> 1165         if s != self._text:
   1166             self._text = str(s)
   1167             self.stale = True

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Answer 1

Version Issue: I was able to replicate with seaborn == 0.10.1, pandas == 1.1.4. But after updating to seaborn == 0.11.1, pandas == 1.2.2, I no longer get the error (I get the second plot below, hitch free). The change is related to seaborn, see the patch notes for 0.11.0 within this section:

relplot() now has the same flexibility as the axes-level functions to accept data in long- or wide-format and to accept data vectors (rather than named variables) in long-form mode.

So updating is a good, simple solution! But if that isn't an option, original answer below.

---

It seems the issue is you are trying to pass actual data to the x and y arguments, rather than column names in data. As seen in examples, the idiomatic way of plotting with seaborn is to name a DataFrame/array with data and then refer to columns in it with x, y, hue, etc. This is a little different from matplotlib where you typically pass the actual iterable data to the plotting function.

So I think the easiest fix is to move your index to a new column, and then plot:

toyData3 = toyData3.reset_index() # creates a new column "index"
sns.relplot(data=toyData3, x='index', y='mean',  kind='line')

Creating the following without errors:

What highlights the issue a little more is that (in my IDE at least) your code creates the following plot before exiting:

Data looks good, but the axis labels were not drawn. So it looks like seaborn is looking for an axis label, which would usually be provided via x and y, but in doing so runs into the error (more specifically, either the x or y is being compared to another value with a ==, returning an array of boolean values rather than a single True or False).