如何使用 ax.legend() 为条形图的子图添加公共图例

Question

不幸的是，我已经尝试过我发现的不同的东西。如何在图的顶部为我的三个子图创建一个共同的图例？下面的代码不起作用，执行被卡住了，下面你还会发现整个代码sn-p。其实我以为我自己能搞定，但我确实需要你的帮助。

fig.tight_layout()
handles, labels = axs[0].get_legend_handles_labels()
fig.legend(handles, labels, fontsize=fs, loc='upper center', bbox_to_anchor=(0.5, 1.05), ncol=4)

这是我的整个代码 sn-p:

df = pd.read_csv('energy_production_ver4.csv', sep=",")

# Figure Properties
fs = 4  # 30
lw = 2  # 3
width_bars=0.5
ec = 'dimgray'

# Create Subplots
fig, axs = plt.subplots(3, sharex=True, sharey=True, num=None, figsize=(25, 16), dpi=300, facecolor='w',
                        edgecolor='k')  # 26 15
plt.rc('text', usetex=True)
plt.rc('font', family='serif')
plt.rc('font', size=fs-1)
plt.gcf().subplots_adjust(bottom=0.15)

#adjust Time Stamp to desired format

temp_x=df["DateTimeStamp"]
y1 = df['Production']
y2 = df['Consumption']
y3 = df['Production CHP 22kW']
y4 = df['Battery charge']
y5 = df['Battery discharge']
y6 = df['Grid supply']
y7 = df['Feed in grid']

# axs[0].grid(True, linestyle=':')
axs[0].yaxis.grid(linestyle=':')
axs[0].tick_params(axis='both', labelsize=fs)
axs[0].bar(temp_x, y1, label=r'$PV^\mathrm{s}$', color='blue', linewidth=lw / 2, width=width_bars)
axs[0].bar(temp_x, y2, label=r'$EV^\mathrm{s}$', color='orange', linewidth=lw / 2, width=width_bars)
axs[0].bar(temp_x, y3, label=r'CHP', color='green', linewidth=lw / 2, width=width_bars)
axs[0].set_ylabel(r'Energy in MWh', fontsize=fs)


axs[1].yaxis.grid(linestyle=':')
axs[1].tick_params(axis='both', labelsize=fs)
axs[1].bar(temp_x, y4, label=r'$BSS^\mathrm{CH}$', color='red', linewidth=lw / 2, width=width_bars)
axs[1].bar(temp_x, y5, label=r'$BSS^\mathrm{d}$', color='purple', linewidth=lw / 2, width=width_bars)
axs[1].set_ylabel(r'Energy in MWh', fontsize=fs)


axs[2].yaxis.grid(linestyle=':')
axs[2].tick_params(axis='both', labelsize=fs)
axs[2].bar(temp_x, y6, label=r'$GRID^\mathrm{CH}$', color='brown', linewidth=lw / 2, width=width_bars)
axs[2].bar(temp_x, y7, label=r'$GRID^\mathrm{d}$', color='pink', linewidth=lw / 2, width=width_bars)
axs[2].set_ylabel(r'Energy in MWh', fontsize=fs)



plt.xticks(rotation=90)
handles, labels = axs.get_legend_handles_labels() #does not work?
fig.legend(handles, labels, fontsize=fs, loc='upper center', bbox_to_anchor=(0.5, 1.05), ncol=4)

plt.show()

Answer 1

总体思路有效（我相信this post 或this one 是您的来源或适用于一个轴）。这里的问题是你有几个轴包裹在axs（这是一个numpy数组）中，所以当你做你写的事情时，你会得到类似的东西

AttributeError: 'numpy.ndarray' object has no attribute 'get_legend_handles_labels'

解决这个问题的方法是遍历所有轴并连接图例（它们的标签和句柄），例如。通过以下方式：

...
handles,labels=[],[]
for ax in axs.flatten():
    h, l = ax.get_legend_handles_labels()
    handles+=h
    labels+=l

fig.legend(handles, labels, fontsize=fs, loc='upper center', bbox_to_anchor=(0.5, 1.05), ncol=4)