Scatter requires x column to be numeric

In [30]: df = pd.DataFrame(dict(a=['A', 'B', 'C'], b=[2, 3, 4]))

In [31]: plt.scatter(df['a'], df['b'])
Out[31]: <matplotlib.collections.PathCollection at 0x7f17e17b41d0>

This works, and produces this:

On the other hand, this doesn’t:

In [32]: df.plot.scatter(x='a', y='b')
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/data/documents/uni/phd/projects/FluxnetTrafficLights/scripts/plots/predictability_plots.py in <module>()                                                                                                     
----> 1 df.plot.scatter(x='a', y='b')

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in scatter(self, x, y, s, c, **kwds)                                                                                           
   2803         axes : matplotlib.AxesSubplot or np.array of them
   2804         """
-> 2805         return self(kind='scatter', x=x, y=y, c=c, s=s, **kwds)
   2806 
   2807     def hexbin(self, x, y, C=None, reduce_C_function=None, gridsize=None,

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in __call__(self, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)                                                                                           
   2625                           fontsize=fontsize, colormap=colormap, table=table,
   2626                           yerr=yerr, xerr=xerr, secondary_y=secondary_y,
-> 2627                           sort_columns=sort_columns, **kwds)
   2628     __call__.__doc__ = plot_frame.__doc__
   2629 

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in plot_frame(data, x, y, kind, ax, subplots, sharex, sharey, layout, figsize, use_index, title, grid, legend, style, logx, logy, loglog, xticks, yticks, xlim, ylim, rot, fontsize, colormap, table, yerr, xerr, secondary_y, sort_columns, **kwds)                                                                                         
   1867                  yerr=yerr, xerr=xerr,
   1868                  secondary_y=secondary_y, sort_columns=sort_columns,
-> 1869                  **kwds)
   1870 
   1871 

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in _plot(data, x, y, subplots, ax, kind, **kwds)
   1650         if isinstance(data, DataFrame):
   1651             plot_obj = klass(data, x=x, y=y, subplots=subplots, ax=ax,
-> 1652                              kind=kind, **kwds)
   1653         else:
   1654             raise ValueError("plot kind %r can only be used for data frames"

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in __init__(self, data, x, y, s, c, **kwargs)
    808             # the handling of this argument later
    809             s = 20
--> 810         super(ScatterPlot, self).__init__(data, x, y, s=s, **kwargs)
    811         if is_integer(c) and not self.data.columns.holds_integer():
    812             c = self.data.columns[c]

~/miniconda3/envs/science/lib/python3.6/site-packages/pandas/plotting/_core.py in __init__(self, data, x, y, **kwargs)
    783             y = self.data.columns[y]
    784         if len(self.data[x]._get_numeric_data()) == 0:
--> 785             raise ValueError(self._kind + ' requires x column to be numeric')
    786         if len(self.data[y]._get_numeric_data()) == 0:
    787             raise ValueError(self._kind + ' requires y column to be numeric')

ValueError: scatter requires x column to be numeric

Why does pandas require x to be numeric if matplotlib doesn’t?

using versions from conda, on Kubuntu 17.10:

matplotlib                2.1.0            py36hba5de38_0  
pandas                    0.20.3                   py36_0  
python                    3.6.3                h0ef2715_3  

[ckaldemeyer]

I have the same error and it occurs only when there is a column of type DateTime which is not even selected for the scatter plot.

[RahulRAhhy123]

I am also getting a same error ValueError: scatter requires x column to be numeric when I am trying plot a column which is of datetime type…how to resolve this @naught101 @ckaldemeyer @sinhrks @bobhaffner @jadolfbr