英文:
Is there a way to suppress legend entry when plotting directly from pandas?
问题
我需要在Matplotlib轴ax上执行多个绘图操作。数据存储在一个pandas DataFrame中，因此自然要使用多个绘图命令：
df.plot(..., ax=ax)
但对于其中一些情况，我不想在图例中显示标签。我尝试添加label="nolegend"，但这并不起作用。
我找到的唯一解决方法是首先从DataFrame中提取数据，然后在循环中使用单独的ax.plot命令绘制它们，并在需要时添加label参数到图例中。这很繁琐。
注意：我不是在问如何完全删除图例。
编辑：在@Vitalizzare发布了他的不错回答之后（谢谢！）：当然，我尝试过了，因为这在文档中提到过，但在我的情况下它不起作用。很难理解为什么，但我认为它与绘制的曲线数量有关。考虑以下代码：
dg = pd.DataFrame(data=np.random.rand(400).reshape(10,40))
fig1, ax = plt.subplots(figsize=(12, 4))
dg.iloc[1:8].plot(ax=ax,style='.-',color='lightgray',legend=False)
dg[[32,33]].iloc[1:8].plot(ax=ax)
这里发生了一些非常奇怪的事情。我得到了不一致的结果！有时候我得到的图没有图例。我不知道为什么。我正在运行mpl 3.5.1和pandas 1.4.2。在我的原始笔记本中，我无法获得正确的行为。
英文:
I need to do multiple plot operations on a Matplotlib axis ax. The data are in a pandas DataFrame, so it is natural to use multiple plotting commands:
df.plot(..., ax=ax)
But for some of those, I don't want a label entry in the legend. I have tried to add label="_nolegend_" but that does not work.
The only workaround I found is to first extract the data from the DataFrame, then plot them in a loop with individual ax.plot commands, adding a label parameter where I want it in the legend. This is tedious.
NOTE: I am not asking how to remove the legend altogether.
EDIT: After @Vitalizzare posted his nice answer (thank you!): Of course, I tried that because it's in the documentation, but it does not work in my case. It isn't easy to understand why, but I think it has to do with the number of curves plotted. Consider this code:
dg = pd.DataFrame(data=np.random.rand(400).reshape(10,40))
fig1, ax = plt.subplots(figsize=(12, 4))
dg.iloc[1:8].plot(ax=ax,style='.-',color='lightgray',legend=False)
dg[[32,33]].iloc[1:8].plot(ax=ax)
Something very strange happens here. I get inconsistent results! Sometimes I get a figure with NO legend. I have no idea why. I am running mpl 3.5.1 and pandas 1.4.2. In my original notebook, I cannot get the correct behavior.
答案1
得分: 1
为了避免绘制图例，我们可以将legend=False作为参数传递给df.plot。要将图例限制为前N个标签，我们可以运行ax.legend(list_of_labels_to_show)。要显示任意一组标签，我们可以在同一坐标轴上分两步绘制数据，对于不应显示的部分，使用legend=False。
示例：
df = pd.DataFrame({
'x': [1,2],
'y': [2,1],
'z': [1.5,1.5]
})
fig, (ax1, ax2, ax3, ax4) = plt.subplots(1, 4, subplot_kw={'aspect':'equal'}, figsize=(12, 3))
fig.suptitle('绘制数据示例')
ax1.set_title('完整图例')
df.plot(ax=ax1)
ax2.set_title('前N个图例')
df.plot(ax=ax2)
ax2.legend(df.columns[:2])
ax3.set_title('选定的图例')
df[['x','z']].plot(ax=ax3)
df['y'].plot(ax=ax3, legend=False)
ax4.set_title('无图例')
df.plot(ax=ax4, legend=False)
fig.savefig('legend_example.png')
输出：
英文:
To avoid plotting a legend we can pass legend=False as a parameter to df.plot. To restrict the legend to the first N labels, we can run ax.legend(list_of_labels_to_show). To show an arbitrary set of labels, we can plot the data separately in 2 steps on the same axes with legend=False on those which shouldn't be displayed.
Example:
df = pd.DataFrame({
'x': [1,2],
'y': [2,1],
'z': [1.5,1.5]
})
fig, (ax1, ax2, ax3, ax4) = plt.subplots(1, 4, subplot_kw={'aspect':'equal'}, figsize=(12, 3))
fig.suptitle('Plotting data with')
ax1.set_title('full legend')
df.plot(ax=ax1)
ax2.set_title('first n legend')
df.plot(ax=ax2)
ax2.legend(df.columns[:2])
ax3.set_title('selected legend')
df[['x','z']].plot(ax=ax3)
df['y'].plot(ax=ax3, legend=False)
ax4.set_title('no legend')
df.plot(ax=ax4, legend=False)
fig.savefig('legend_example.png')
Output:
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论