使用y选项时将图例添加到pandas.dataframe.plot

Question

If I plot a DataFrame as

df.plot()

then each line appears in the legend identifying which column it corresponds to, as it should be. However, if I plot it as

df.plot(x=0, y=[2,3])

then everything is plotted normally, however there is no legend identifying which line is which column.

I have tried

df.plot(x=0, y=[2,3], label=['2','3'])
df.plot(x=0, y=[2,3], legend=['2','3'])

but nothing works. The only workaround is to set

plt.legend(['2','3'])

afterwards, but I am not sure that the order of the legend list is the same order that pandas uses for plotting, so I don't know if the legend really matches the lines. Is there a way to make pandas plot the legend is this case?

I am using pandas 0.14.

Answer 1

As @unutbu writes, this works well for current versions of Pandas. On older versions, however, I found it easier just to "narrow down" the DataFrame to the relevant columns.

Eg, suppose you have

df = pd.DataFrame({'a': [10, 2], 'b': [2, 3], 'c': [1, 3], 'd': [4, 5]})

Then to plot 'b' and 'c' as a function of 'a' :

df[['a', 'b', 'c']].plot(x='a');

This gives you the legend ordering for free.