ENH: Let DataFrame.plot(kind="scatter") scatter plot each column against index if x/y not given.

[randolf-scholz]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

To my surprise, df.plot(kind="scatter") simply raises a ValueError complaining about missing "x" and "y".
In many cases, particularly time series data, the sensible thing to do is to scatter plot each channel against the index.

Feature Description

When x/y is not given scatter plot columns against the index.

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

x = np.arange(0, 10, 0.1)
df = pd.DataFrame({"sin": np.sin(x), "cos": np.cos(x)}, index=x)

try:  # without subplots
    df.plot(kind="scatter")
except ValueError as e:
    print(e)
    fig, ax = plt.subplots()
    for col in df:
        ax.scatter(df.index, df[col], label=col)
    ax.legend()
    plt.show()

try:  # with subplots
    df.plot(subplots=True, kind="scatter")
except ValueError as e:
    print(e)
    fig, axes = plt.subplots(len(df.columns), 1)
    for ax, col in zip(axes, df):
        ax.scatter(df.index, df[col], label=col)
        ax.legend()
    plt.show()

Alternative Solutions

Of course, one could just do it manually.

Additional Context

No response

[luke396]

The bug does exist in the current main branch.

Should we add default values (e.g., index for x) for x and y when the user does not input them? Currently, the code does not attempt to provide default values, so None is returned, which raises this error.

pandas/pandas/plotting/_core.py

Lines 890 to 898 in f8a37a7

	if backend_name == "pandas.plotting._matplotlib":
	kwargs = dict(arg_def, **pos_args, **kwargs)
	else:
	kwargs = dict(pos_args, **kwargs)
	x = kwargs.pop("x", None)
	y = kwargs.pop("y", None)
	kind = kwargs.pop("kind", "line")
	return x, y, kind, kwargs

Also, maybe unrelatedly, it seems that kind='scatter' works just fine when inputting a Dataframe. Should we expand this to Series by setting x as s.index?

pandas/pandas/plotting/_core.py

Line 792 in f8a37a7

_dataframe_kinds = ("scatter", "hexbin")

pandas/pandas/plotting/_core.py

Lines 924 to 928 in f8a37a7

	if kind in self._dataframe_kinds:
	if isinstance(data, ABCDataFrame):
	return plot_backend.plot(data, x=x, y=y, kind=kind, **kwargs)
	else:
	raise ValueError(f"plot kind {kind} can only be used for data frames")

[luke396]

Sorry for my rushed reply. After a deeper view, adding something for the ScatterPlot class may be better for this issue.

pandas/pandas/plotting/_matplotlib/core.py

Line 1187 in 74e8c00

class ScatterPlot(PlanePlot):

[joshdunnlime]

Allowing scatter to follow the df.plot() notation and functionality seems sensible to me. That is to say, using the index as the default x-axis and plotting all columns as y values, with the first column taking the first colour in the colour palette and the second column taking the colour of the second colour in the colour palette.

This would intuitively allow users to switch from: df.plot() to df.plot.scatter(), without changing any kwargs, when realising there data is not order or not sequential.