How to calculate sum and average bars for every histogram bin

Question

I would like to plot stacked bar charts. The first 3 bars are red, black and blue as shown below.

I want to add a fourth bar that is the 'sum' of the values of red, black and blue bars.

(In another case I want to add a fourth bar that is the 'average' of red, black and blue bars.)

For example, this is my code:

import numpy as np
import matplotlib.pyplot as plt

np.random.seed(19680801)

n_bins = 20
x = np.random.randn(1000, 3)

fig, ax0 = plt.subplots(nrows=1, ncols=1)

colors = ['red', 'black', 'blue']
ax0.hist(x, n_bins, density=True, histtype='bar', color=colors, label=colors)
ax0.legend(prop={'size': 10})
ax0.set_title('bars with legend')



fig.tight_layout()
plt.show()

This figure is without the 4th bar.

Any suggestions how of how I can show the 4th bar in this stack bar chart?

Answer 1

Use np.histogram to calculate hist and bin_edges for each group, calculate the sum and mean of each group, and plot bar plots with the absolute bar height for each bin.

The easiest way to plot grouped, or stacked bars, is from a pandas.DataFrame with pandas.DataFrame.plot.

Tested in python 3.11.2, pandas 2.0.1, matplotlib 3.7.1, numpy 1.24.3

import numpy as np
import pandas as pd

# data
np.random.seed(19680801)
n_bins = 20
x = np.random.randn(1000, 3)

# calculate bin_edges for the combined values
_, be = np.histogram(x, bins=n_bins, density=True)

# calculate hist for each sample
data = {f's{i}': np.histogram(col, bins=be, density=True)[0] for i, col in enumerate(x.T)}

# round the values of the bin edges to be used as xtick labels in the plot
be = be.round(1)

# stack the arrays for each bin group
groups = np.column_stack(list(data.values()))

# calculate the total
data['tot'] = groups.sum(axis=1)

# calculate the mean
data['mean'] = groups.mean(axis=1)

# create a dataframe from data
df = pd.DataFrame(data=data)

# plot 
ax = df.plot(kind='bar', width=0.85, ec='k', figsize=(11, 6), rot=0)

# update the xticks by shifting them to the left by 0.5, and updating the labels
ax.set_xticks(ticks=np.arange(0, len(be))-0.5, labels=be)

# add some cosmetics
ax.grid(axis='x')
ax.spines[['right', 'top']].set_visible(False)
_ = ax.legend(bbox_to_anchor=(1, 0.5), loc='center left', frameon=False)

• Use .iloc, or y=['s0', 's1', 's2', 'mean'] to select the specific columns to plot.

# plot 
ax = df.iloc[:, [0, 1, 2, 4]].plot(kind='bar', width=0.85, ec='k', figsize=(14, 6), rot=0)

# add some cosmetics
ax.set_xticks(ticks=np.arange(0, len(be))-0.5, labels=be)
ax.grid(axis='x')
ax.spines[['right', 'top']].set_visible(False)
_ = ax.legend(bbox_to_anchor=(1, 0.5), loc='center left', frameon=False)

# add some labels
for c in ax.containers:
    ax.bar_label(c, fmt=lambda x: f'{x:0.3f}' if x > 0.005 else '', label_type='edge', fontsize=6, rotation=90, padding=3)
    
ax.margins(y=0.15)