如何修复 matplotlib 中所有子图的图例和主标题？

Question

我是 Python 新手，我一直在使用虚拟数据集来练习 Python。以前我在生成子图、然后绘制频率和比例 % 时遇到了麻烦，但现在我今天已经克服了它们。现在，我正在努力修复一些装饰性的东西，尤其是传说和情节标题。

这是生成整个虚拟数据集的可重现代码：

d = {
    'SeniorCitizen': [0,1,0,0,0,0,0,1,0,1,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0] , 
    'CollegeDegree': [0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,1,1,1,1] , 
    'Married': [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1] , 
    'FulltimeJob': [1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,1,1,0,0,1,1,0,0,0,1] , 
    'DistancefromBranch': [7,9,14,20,21,12,22,25,9,9,9,12,13,14,16,25,27,4,14,14,20,19,15,23,2] , 
    'ReversedPayment': [0,0,0,0,1,0,1,0,0,0,0,0,0,0,0,0,1,0,1,0,0,1,0,1,0] }
CarWash = pd.DataFrame(data = d)


categoricals = ['SeniorCitizen','CollegeDegree','Married','FulltimeJob','ReversedPayment']
        numerical = ['DistancefromBranch']
CarWash[categoricals] = CarWash[categoricals].astype('category')

下面是我尝试生成频率和比例%，并列比较：

plt.suptitle("Distribution of target variable across the categorical variables - frequencies # and proportions %")

nrow = 1
ncol = len(categoricals[:-1])
figure, axes = plt.subplots(nrow,ncol, figsize = (40,10))
for i,ax in zip(categoricals[:-1],axes.flatten()):
    
    # plots frequencies
    CarWash.groupby([i,'ReversedPayment']).size().reset_index().pivot(index = i,columns = 'ReversedPayment').plot(kind = 'bar', stacked = True, ax = ax, sharey=True)
    ax.tick_params(axis='both', labelsize = 25, labelrotation = 0)   
    ax.set_title(i,fontsize = 30)
    ax.set_xlabel("")
    #ax.get_legend().remove()
    # labels data
    for p in ax.patches:       
        x_adjust = 0.25
        
        value = p.get_height()
        X = p.get_x() + x_adjust
        Y = p.get_y() + p.get_height()/2
        
        XY = (X,Y)
        if value != 0:
            ax.annotate(int(value),XY,fontsize = 25)


nrow = 1
ncol = len(categoricals[:-1])
figure, axes = plt.subplots(nrow,ncol, figsize = (40,10))
for i,ax in zip(categoricals[:-1],axes.flatten()):
    
    # plots proportions    
    CarWash.groupby([i,'ReversedPayment']).size().reset_index().pivot(index = i, columns = 'ReversedPayment').apply(lambda x: x/x.sum(),axis=1).plot(kind = 'bar', stacked = True, ax = ax)
    ax.tick_params(axis='both', labelsize = 25, labelrotation = 0)   
    ax.set_title(i,fontsize = 30)
    ax.set_xlabel("")
    #ax.get_legend().remove()
    # labels data
    for p in ax.patches:       
        x_adjust = 0.25/3
        
        value = p.get_height()
        X = p.get_x() + x_adjust
        Y = p.get_y() + p.get_height()/2
        
        XY = (X,Y)
        if value != 0:
            ax.annotate(str(round(value*100,1)) + "%",XY,fontsize = 25)

下面是我的输出：

所以，我对化妆品编码的问题是：

1. 标题：我尝试使用plt.suptitle() 引入一个主标题，但这并没有达到预期效果（我看不到任何输出）。还尝试了其他方法，但其他所有方法都引发了错误。
2. Legend：Legend 看起来很难看，我不需要它们用于所有子情节。我正在尝试为所有子图只获取一个图例，这将非常棒并且可以节省空间。我尝试了类似 plt.legend([CarWash[i],CarWash['ReversedPayment']], ['Blue', 'Orange']) 的方法，但没有成功。

欢迎和非常感谢任何 cmets / 建议。谢谢。

Answer 1

您在为绘图创建图形之前执行plt.suptitle，这就是标题未出现在最终绘图中的原因。有关如何为多个子图创建公共图例的答案，请参阅 here。综合考虑，您可以尝试以下方法：

nrow = 2
ncol = len(categoricals[:-1])
fig, axes = plt.subplots(nrow,ncol, figsize = (40,20))
plt.suptitle("Distribution of target variable across the categorical variables - frequencies # and proportions %", size=40, y=1.01)

for i,ax in zip(categoricals[:-1],axes.flatten()):
    
    # plots frequencies
    df = CarWash[[i,'ReversedPayment']].value_counts().unstack().fillna(0)
    ax.bar(df.index, df[0], label='0')
    ax.bar(df.index, df[1], bottom=df[0], label='1')
    ax.tick_params(axis='both', labelsize = 25, labelrotation = 0)   
    ax.set_title(i,fontsize = 30, y=1.05)
    ax.set_xticks([0,1])
    
    # labels data
    for p in ax.patches:       
        x_adjust = 0.35
        
        value = p.get_height()
        X = p.get_x() + x_adjust
        Y = p.get_y() + p.get_height()/2
        
        XY = (X,Y)
        if value != 0:
            ax.annotate(int(value),XY,fontsize = 25)
                       
for i,ax in zip(categoricals[:-1],axes.flatten()[ncol:]):
    
    # plots proportions
    df = CarWash[[i,'ReversedPayment']].value_counts().unstack().fillna(0)
    df = (df.T/df.T.sum()).T*100
    ax.bar(df.index, df[0], label='0')
    ax.bar(df.index, df[1], bottom=df[0], label='1')
    ax.tick_params(axis='both', labelsize = 25, labelrotation = 0)   
    ax.set_title(i,fontsize = 30, y=1.05)
    ax.set_xticks([0,1])
    ax.set_ylim(0,110)

    # labels data
    for p in ax.patches:       
        x_adjust = 0.25
        
        value = p.get_height()
        X = p.get_x() + x_adjust
        Y = p.get_y() + p.get_height()/2
        
        XY = (X,Y)
        if value != 0:
            ax.annotate(str(round(value,1)) + "%",XY,fontsize = 25)
            
plt.subplots_adjust(hspace=0.4)

# plot legend
handles, labels = ax.get_legend_handles_labels()
fig.legend(handles, labels, loc=(0.95, 0.1), prop={'size': 30})
plt.show()

它给出：

Answer 2

也许您可以将标题移动到循环中并在迭代后添加图例，如下所示。

nrow = 1
ncol = len(categoricals[:-1])
figure, axes = plt.subplots(nrow,ncol, figsize = (20,5))


for i,ax in zip(categoricals[:-1],axes.flatten()):    
    # plots frequencies
    plt.suptitle("Distribution of target variable across the categorical variables - frequencies # and proportions %")
CarWash.groupby([i,'ReversedPayment']).size().reset_index().pivot(index = i,columns = 'ReversedPayment').plot(kind = 'bar', stacked = True, ax = ax, sharey=True)
    ax.tick_params(axis='both', labelsize = 15, labelrotation = 0)   
    ax.set_title(i,fontsize = 15)
    ax.set_xlabel("")
    legend = ax.legend()
    #ax.get_legend().remove()
    # labels data
    for p in ax.patches:       
    x_adjust = 0.25
    
        value = p.get_height()
        X = p.get_x() + x_adjust
        Y = p.get_y() + p.get_height()/2
        legend.remove()
    
        XY = (X,Y)
        if value != 0:
        ax.annotate(int(value),XY,fontsize = 15)    

ax.legend(loc=2, bbox_to_anchor=(-4.2,1.1),fontsize=10) 

输出图如下所示，仍然需要进行一些调整以包括第二行子图和定位/尺寸。

对于涉及许多标签、图例、附录的科学情节，您可以使用另一个 Python 驱动的程序，称为 Veusz。这也很容易使用。 Veusz website