python - Seaborn Barplot - Displaying Values

52
votes

I'm looking to see how to do two things in Seaborn with using a bar chart to display values that are in the dataframe, but not in the graph

1) I'm looking to display the values of one field in a dataframe while graphing another. For example, below, I'm graphing 'tip', but I would like to place the value of 'total_bill' centered above each of the bars (i.e.325.88 above Friday, 1778.40 above Saturday, etc.)

2) Is there a way to scale the colors of the bars, with the lowest value of 'total_bill' having the lightest color (in this case Friday) and the highest value of 'total_bill' having the darkest. Obviously, I'd stick with one color (i.e. blue) when I do the scaling.

Thanks! I'm sure this is easy, but i'm missing it..

While I see that others think that this is a duplicate of another problem (or two), I am missing the part of how I use a value that is not in the graph as the basis for the label or the shading. How do I say, use total_bill as the basis. I'm sorry, but I just can't figure it out based on those answers.

Starting with the following code,

import pandas as pd
import seaborn as sns
%matplotlib inline
df=pd.read_csv("https://raw.githubusercontent.com/wesm/pydata-    book/master/ch08/tips.csv", sep=',')
groupedvalues=df.groupby('day').sum().reset_index()
g=sns.barplot(x='day',y='tip',data=groupedvalues)

I get the following result:

enter image description here

Interim Solution:

for index, row in groupedvalues.iterrows():
    g.text(row.name,row.tip, round(row.total_bill,2), color='black', ha="center")

enter image description here

On the shading, using the example below, I tried the following:

import pandas as pd
import seaborn as sns
%matplotlib inline
df=pd.read_csv("https://raw.githubusercontent.com/wesm/pydata-book/master/ch08/tips.csv", sep=',')
groupedvalues=df.groupby('day').sum().reset_index()

pal = sns.color_palette("Greens_d", len(data))
rank = groupedvalues.argsort().argsort() 
g=sns.barplot(x='day',y='tip',data=groupedvalues)

for index, row in groupedvalues.iterrows():
    g.text(row.name,row.tip, round(row.total_bill,2), color='black', ha="center")

But that gave me the following error:

AttributeError: 'DataFrame' object has no attribute 'argsort'

So I tried a modification:

import pandas as pd
import seaborn as sns
%matplotlib inline
df=pd.read_csv("https://raw.githubusercontent.com/wesm/pydata-book/master/ch08/tips.csv", sep=',')
groupedvalues=df.groupby('day').sum().reset_index()

pal = sns.color_palette("Greens_d", len(data))
rank=groupedvalues['total_bill'].rank(ascending=True)
g=sns.barplot(x='day',y='tip',data=groupedvalues,palette=np.array(pal[::-1])[rank])

and that leaves me with

IndexError: index 4 is out of bounds for axis 0 with size 4

8
pythonpandasmatplotlibseaborn
matplotlib.org/users/annotations.html - Paul H
Did you actually seach for a solution before posting the question? Here is a way to set labels on bars and here is a way to change the bar color. If you have problems implementing either solution you can ask a specific question about it. - ImportanceOfBeingErnest
Trying to work through the examples again.. i'll come back.. ty. If i figure it out, will post answer - Stumbling Through Data Science
I put one possible idea out there for labels.. shading still escapes me - Stumbling Through Data Science
Added a lot of detail and code above.. cant seem to get further.. thx - Stumbling Through Data Science

8 Answers

53
votes

Let's stick to the solution from the linked question (Changing color scale in seaborn bar plot). You want to use argsort to determine the order of the colors to use for colorizing the bars. In the linked question argsort is applied to a Series object, which works fine, while here you have a DataFrame. So you need to select one column of that DataFrame to apply argsort on.

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

df = sns.load_dataset("tips")
groupedvalues=df.groupby('day').sum().reset_index()

pal = sns.color_palette("Greens_d", len(groupedvalues))
rank = groupedvalues["total_bill"].argsort().argsort() 
g=sns.barplot(x='day',y='tip',data=groupedvalues, palette=np.array(pal[::-1])[rank])

for index, row in groupedvalues.iterrows():
    g.text(row.name,row.tip, round(row.total_bill,2), color='black', ha="center")

plt.show()

enter image description here

rank()1int
rank = groupedvalues['total_bill'].rank(ascending=True).values
rank = (rank-1).astype(np.int)
60
votes

Works with single ax or with matrix of ax (subplots)

from matplotlib import pyplot as plt
import numpy as np

def show_values_on_bars(axs):
    def _show_on_single_plot(ax):        
        for p in ax.patches:
            _x = p.get_x() + p.get_width() / 2
            _y = p.get_y() + p.get_height()
            value = '{:.2f}'.format(p.get_height())
            ax.text(_x, _y, value, ha="center") 

    if isinstance(axs, np.ndarray):
        for idx, ax in np.ndenumerate(axs):
            _show_on_single_plot(ax)
    else:
        _show_on_single_plot(axs)

fig, ax = plt.subplots(1, 2)
show_values_on_bars(ax)
40
votes

Just in case if anyone is interested in labeling horizontal barplot graph, I modified Sharon's answer as below: 

def show_values_on_bars(axs, h_v="v", space=0.4):
    def _show_on_single_plot(ax):
        if h_v == "v":
            for p in ax.patches:
                _x = p.get_x() + p.get_width() / 2
                _y = p.get_y() + p.get_height()
                value = int(p.get_height())
                ax.text(_x, _y, value, ha="center") 
        elif h_v == "h":
            for p in ax.patches:
                _x = p.get_x() + p.get_width() + float(space)
                _y = p.get_y() + p.get_height()
                value = int(p.get_width())
                ax.text(_x, _y, value, ha="left")

    if isinstance(axs, np.ndarray):
        for idx, ax in np.ndenumerate(axs):
            _show_on_single_plot(ax)
    else:
        _show_on_single_plot(axs)

Two parameters explained:

h_v - Whether the barplot is horizontal or vertical. "h" represents the horizontal barplot, "v" represents the vertical barplot.

space - The space between value text and the top edge of the bar. Only works for horizontal mode.

Example:

show_values_on_bars(sns_t, "h", 0.3)

enter image description here

12
votes
plt.figure(figsize=(15,10))
graph = sns.barplot(x='name_column_x_axis', y="name_column_x_axis", data = dataframe_name ,  color="salmon")
for p in graph.patches:
        graph.annotate('{:.0f}'.format(p.get_height()), (p.get_x()+0.3, p.get_height()),
                    ha='center', va='bottom',
                    color= 'black')
2
votes

Hope this helps for item #2: a) You can sort by total bill then reset the index to this column b) Use palette="Blue" to use this color to scale your chart from light blue to dark blue (if dark blue to light blue then use palette="Blues_d")

import pandas as pd
import seaborn as sns
%matplotlib inline

df=pd.read_csv("https://raw.githubusercontent.com/wesm/pydata-book/master/ch08/tips.csv", sep=',')
groupedvalues=df.groupby('day').sum().reset_index()
groupedvalues=groupedvalues.sort_values('total_bill').reset_index()
g=sns.barplot(x='day',y='tip',data=groupedvalues, palette="Blues")
1
votes

A simple way to do so is to add the below code (for Seaborn):

for p in splot.patches:
    splot.annotate(format(p.get_height(), '.1f'), 
                   (p.get_x() + p.get_width() / 2., p.get_height()), 
                   ha = 'center', va = 'center', 
                   xytext = (0, 9), 
                   textcoords = 'offset points')

Example :

splot = sns.barplot(df['X'], df['Y'])
# Annotate the bars in plot
for p in splot.patches:
    splot.annotate(format(p.get_height(), '.1f'), 
                   (p.get_x() + p.get_width() / 2., p.get_height()), 
                   ha = 'center', va = 'center', 
                   xytext = (0, 9), 
                   textcoords = 'offset points')    
plt.show()
0
votes
import seaborn as sns

fig = plt.figure(figsize = (12, 8))
ax = plt.subplot(111)

ax = sns.barplot(x="Knowledge_type", y="Percentage", hue="Distance", data=knowledge)

for p in ax.patches:
    ax.annotate(format(p.get_height(), '.2f'), (p.get_x() + p.get_width() / 2., p.get_height()), 
       ha = 'center', va = 'center', xytext = (0, 10), textcoords = 'offset points')
0
votes

New in matplotlib 3.4.0

There is now a built-in Axes.bar_label to automatically label the bar containers, so labeling now becomes a one-liner:

ax = sns.barplot(x='day', y='tip', data=groupedvalues)
ax.bar_label(ax.containers[0])

And just for completeness, for the color-ranked version, here is a slightly modified version of Ernest's code using Series.rank with Series.sub:

rank = groupedvalues.total_bill.rank().sub(1).astype(int)
palette = np.array(sns.color_palette('Blues_d', len(rank)))[rank]

ax = sns.barplot(x='day', y='tip', palette=palette, data=groupedvalues)
ax.bar_label(ax.containers[0])

sns.barplot labeled and ranked