Adding labels to seaborn bars

Question

I'm trying to create two, vertically aligned, horizontal grouped bar charts. I have a huge amount of data for several Machine Learning models and their corresponding runtimes and would like to display all this data in a meaningful way. My attempt so far looks as follows:

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

labels = ['MLP','FCN','ResNet','ROCKET','1-NN DTW','LightGBM','XGBoost','CatBoost']
            
Data1_Accuracy = [20, 34, 30, 35, 27,77.83125,78.7204167,78.5354167]
Data2_Accuracy = [20, 34, 30, 35, 27,75.7979167,76.2520833,77.87]
Data3_Accuracy = [20, 34, 30, 35, 27,80.14625,81.5033333,81.4625]
Data4_Accuracy = [20, 34, 30, 35, 27,78.3841667,79.34875,80.5270833]
Data5_Accuracy = [20, 34, 30, 35, 27,79.2495833,77.5370833,79.2666667]
Data6_Accuracy = [20, 34, 30, 35, 27,77.03125,77.2429167,77.9960275]
Data7_Accuracy = [20, 34, 30, 35, 27,81.3241667,80.5408333,84.2083333]
Data8_Accuracy = [20, 34, 30, 35, 27,78.1470833,78.1225,80.2754167]
Data9_Accuracy = [20, 34, 30, 35, 27,80.7383333,79.9358333,79.6916667]
Data10_Accuracy = [20, 34, 30, 35, 27,74.1095833,73.0879167,73.0529167]
Data11_Accuracy = [20, 34, 30, 35, 27,78.4775,77.8658333,78.35]
Data12_Accuracy = [20, 34, 30, 35, 27,73.0991667,71.9683333,72.75625]
Data13_Accuracy = [20, 34, 30, 35, 27,79.03,79.575,80.3870833]
Data14_Accuracy = [20, 34, 30, 35, 27,81.0241667,81.455,80.5516667]
Data15_Accuracy = [20, 34, 30, 35, 27,79.4829167,80.01375,81.68]
Data16_Accuracy = [20, 34, 30, 35, 27,81.1158333,80.9795833,80.6541667]

Data1_Times = [20, 34, 30, 35, 27,829.0177925,58.6558111,8493.968922]
Data2_Times = [20, 34, 30, 35, 27,604.5935536,64.3871907,6833.585728]
Data3_Times = [20, 34, 30, 35, 27,1286.01507,92.4329714,6821.308612]
Data4_Times = [20, 34, 30, 35, 27,757.3903304,78.7253731,5455.483287]
Data5_Times = [20, 34, 30, 35, 27,401.3722335,30.4119882,5160.041989]
Data6_Times = [20, 34, 30, 35, 27,321.4673242,54.1971346,4465.557807]
Data7_Times = [20, 34, 30, 35, 27,2598.48826,193.1256487,10811.65574]
Data8_Times = [20, 34, 30, 35, 27,1545.059628,139.9638344,7784.332016]
Data9_Times = [20, 34, 30, 35, 27,663.416329,615.3660963,3560.337827]
Data10_Times = [20, 34, 30, 35, 27,670.1615828,621.8249994,3567.653313]
Data11_Times = [20, 34, 30, 35, 27,619.1959161,572.3292757,3493.582855]
Data12_Times = [20, 34, 30, 35, 27,626.107683,579.0746278,3528.605614]
Data13_Times = [20, 34, 30, 35, 27,2936.5633,2631.284413,6465.254111]
Data14_Times = [20, 34, 30, 35, 27,2967.02757,2672.068268,6551.57865]
Data15_Times = [20, 34, 30, 35, 27,4102.511475,3711.899848,7704.401239]
Data16_Times = [20, 34, 30, 35, 27,4075.485739,3726.896591,7737.482708]

Data1_TimesInHours = np.array(Data1_Times) / 3600
Data2_TimesInHours = np.array(Data2_Times) / 3600
Data3_TimesInHours = np.array(Data3_Times) / 3600
Data4_TimesInHours = np.array(Data4_Times) / 3600
Data5_TimesInHours = np.array(Data5_Times) / 3600
Data6_TimesInHours = np.array(Data6_Times) / 3600
Data7_TimesInHours = np.array(Data7_Times) / 3600
Data8_TimesInHours = np.array(Data8_Times) / 3600
Data9_TimesInHours = np.array(Data9_Times) / 3600
Data10_TimesInHours = np.array(Data10_Times) / 3600
Data11_TimesInHours = np.array(Data11_Times) / 3600
Data12_TimesInHours = np.array(Data12_Times) / 3600
Data13_TimesInHours = np.array(Data13_Times) / 3600
Data14_TimesInHours = np.array(Data14_Times) / 3600
Data15_TimesInHours = np.array(Data15_Times) / 3600
Data16_TimesInHours = np.array(Data16_Times) / 3600

accuraciesDataFrame = pd.DataFrame({'Index': labels,
                   'Data1_Accuracy': Data1_Accuracy,
                   'Data2_Accuracy': Data2_Accuracy,
                   'Data3_Accuracy': Data3_Accuracy,
                   'Data4_Accuracy': Data4_Accuracy,
                   'Data5_Accuracy': Data5_Accuracy,
                   'Data6_Accuracy': Data6_Accuracy,
                   'Data7_Accuracy': Data7_Accuracy,
                   'Data8_Accuracy': Data8_Accuracy,
                   'Data9_Accuracy': Data9_Accuracy,
                   'Data10_Accuracy': Data10_Accuracy,
                   'Data11_Accuracy': Data11_Accuracy,
                   'Data12_Accuracy)': Data12_Accuracy,
                   'Data13_Accuracy': Data13_Accuracy,
                   'Data14_Accuracy': Data14_Accuracy,
                   'Data15_Accuracy': Data15_Accuracy,
                   'Data16_Accuracy': Data16_Accuracy},
                    columns = ['Index','Data1_Accuracy','Data2_Accuracy','Data3_Accuracy','Data4_Accuracy','Data5_Accuracy','Data6_Accuracy','Data7_Accuracy','Data8_Accuracy','Data9_Accuracy','Data10_Accuracy',
                               'Data11_Accuracy','Data12_Accuracy','Data13_Accuracy','Data14_Accuracy','Data15_Accuracy','Data16_Accuracy'])
        
timesDataFrame = pd.DataFrame({'Index': labels,
           'Data1_TimesInHours': Data1_TimesInHours,
           'Data2_TimesInHours': Data2_TimesInHours,
           'Data3_TimesInHours': Data3_TimesInHours,
           'Data4_TimesInHours': Data4_TimesInHours,
           'Data5_TimesInHours': Data5_TimesInHours,
           'Data6_TimesInHours': Data6_TimesInHours,
           'Data7_TimesInHours': Data7_TimesInHours,
           'Data8_TimesInHours': Data8_TimesInHours,
           'Data9_TimesInHours': Data9_TimesInHours,
           'Data10_TimesInHours': Data10_TimesInHours,
           'Data11_TimesInHours': Data11_TimesInHours,
           'Data12_TimesInHours': Data12_TimesInHours,
           'Data13_TimesInHours': Data13_TimesInHours,
           'Data14_TimesInHours': Data14_TimesInHours,
           'Data15_TimesInHours': Data15_TimesInHours,
           'Data16_TimesInHours': Data16_TimesInHours},
           columns = [
               'Index','Data1_TimesInHours','Data2_TimesInHours','Data3_TimesInHours','Data4_TimesInHours',
                       'Data5_TimesInHours','Data6_TimesInHours','Data7_TimesInHours','Data8_TimesInHours','Data9_TimesInHours','Data10_TimesInHours',
                       'Data11_TimesInHours','Data12_TimesInHours','Data13_TimesInHours','Data14_TimesInHours','Data15_TimesInHours','Data16_TimesInHours'
                       ])
 
accuraciesDataFrameMelted = pd.melt(accuraciesDataFrame, id_vars=['Index'])
timesDataFrameMelted = pd.melt(timesDataFrame, id_vars=['Index'])

fig, axs = plt.subplots(1,2)
fig.set_size_inches(30,10)

xRangeFirstChart = list(range(0,101))

fig.suptitle('Rounded accuracies (%) and times for training and evaluation (h) for different data types and models',fontsize=26)

g1 = sns.barplot(x='value', y='Index', hue='variable', data=accuraciesDataFrameMelted, ax=axs[0])

axs[0].set_xlim([xRangeFirstChart[0],xRangeFirstChart[-1]])
axs[0].set_ylabel('Model',fontsize=24)
axs[0].set_xlabel('Rounded Accuracy (%)',fontsize=24)
axs[0].set_title('Rounded accuracies (%) for different data types and models',fontsize=22)

g2 = sns.barplot(x='value', y='Index', hue='variable', data=timesDataFrameMelted, ax=axs[1])

axs[0].get_legend().remove()
axs[1].get_legend().remove()

axs[1].get_yaxis().set_visible(False)
axs[1].set_xlabel('Training and evaluation time (h)',fontsize=24)
axs[1].set_title('Rounded training and evaluation time (h) for different data types and models',fontsize=22)

plt.savefig('PathToFigure/MyFigure.png', dpi=300, bbox_inches='tight', pad_inches=0)

What I'm missing is a way to write the labels "Data 1", "Data 2", Data 3", etc... in every bar. Please refer to the image for a visualization of what I'm trying to achieve. Any help is highly appreciated!

David Erickson · Accepted Answer

Since there are so many bars in one graph, I would use sns.catplot to draw the the different categories into a Facet Grid and then it would be much better for adding labels, which you can do with the custom function add_labels (please note the different parameters -- feel free to remove some/add others. I have adapted from this solution).

You could also make the x-axis more variable if you pass sharex=False when creating the catplots (see end of this solution)

Also, sns.catplot doesn't work well with adding to subplots, so that you can save as one figure. This is why I use plt.close(fig) to get rid of the blank figure we created, and this would also mean adding any formatting (such as adding a title) to that figure would be pointless, since we are getting rid of the figure at the end; however, there are hacks. One is to save as separate figures and use a solution from here: to combine into one .pdf. I think it would be better to have the extra space of one graph per page or image. Another option is to use somewhat of a hack to get into one figure:

fig, ax = plt.subplots(nrows=2)
sns.set_context('paper', font_scale=1.4)
plt.style.use('dark_background')

n_cols=4 #this is used later in a couple of places to make dynamic 
g1 = sns.catplot(data=accuraciesDataFrameMelted, x='value', y='variable', col='Index', kind='bar', 
                 col_wrap=n_cols, ax=ax[0])
g1.fig.suptitle('Rounded accuracies (%) for different data types and models',fontsize=22)      
plt.subplots_adjust(top=0.9, bottom=-0.5)        

g2 = sns.catplot(data=timesDataFrameMelted, x='value', y='variable', col='Index', kind='bar', 
                 col_wrap=n_cols, ax=ax[1])
g2.fig.suptitle('Rounded training and evaluation time (h) for different data types and models',fontsize=22)
plt.subplots_adjust(top=0.9, bottom=-0.5)      

def add_labels(graph, category_size, axis_number, omit_thresh, width_var, num_format):
    for i in range(category_size):
        ax = graph.facet_axis(axis_number,i)
        for p in ax.patches:
            if p.get_width() > omit_thresh: # omit labels close to zero or other threshold
                width = p.get_width() * width_var   # get bar length
                ax.text(width,       # set the text at 1 unit right of the bar
                p.get_y() + p.get_height() / 2, # get Y coordinate + X coordinate / 2
                num_format.format(p.get_width()), # set variable to display, 2 decimals
                ha = 'center',   # horizontal alignment
                va = 'center')  # vertical alignment
            else:
                pass


l1 = len(accuraciesDataFrameMelted['Index'].unique())
l2 = len(timesDataFrame['Index'].unique())
add_labels(graph=g1, category_size=l1, axis_number=0, omit_thresh=1, width_var=0.5, num_format='{:1.0f}')
add_labels(graph=g2, category_size=l2, axis_number=1, omit_thresh=0.1, width_var=0.5, num_format='{:1.2f}')  

for g, i in zip([g1,g2], [0, n_cols]):
    g.axes[i].set_ylabel('Model')     
for g in [g1,g2]:
    g.set_titles("{col_name}", fontsize=12)
g1.set_axis_labels(x_var="Rounded Accuracy (%)", y_var="Model")
g2.set_axis_labels(x_var="Training and evaluation time (h)", y_var="Model")
plt.close(fig)
g1.fig.savefig('g1.pdf',dpi=300, bbox_inches = "tight")
g2.fig.savefig('g2.pdf',dpi=300, bbox_inches = "tight")
plt.show()

(Zoomed In to show first graph)

(Zoomed Out to show both graphs)

You could also make the x-axis more variable if you pass sharex=False when creating the catplot, by making the changes below (pass sharex and change one of the params in my function to `omit_thresh=0:

g1 = sns.catplot(data=accuraciesDataFrameMelted, x='value', y='variable', 
col='Index', kind='bar', 
             col_wrap=n_cols, ax=ax[0], sharex=False)  
g2 = sns.catplot(data=timesDataFrameMelted, x='value', y='variable', col='Index', kind='bar', 
             col_wrap=n_cols, ax=ax[1], sharex=False)
add_labels(graph=g1, category_size=l1, axis_number=0, omit_thresh=0, width_var=0.5, num_format='{:1.0f}')
add_labels(graph=g2, category_size=l2, axis_number=1, omit_thresh=0, width_var=0.5, num_format='{:1.3f}')

Adding labels to seaborn bars

Answers (1)

Related Questions