Plot two histograms on single chart

Planning your histograms

Good histograms require selecting the right bin sizes, using distinguishable colors, and properly normalizing your data, especially when you're comparing datasets.

Bin edge consistency

Using the same bin edges makes the comparison between the histograms clear and meaningful:

import numpy as np

# Define your bins
bins = np.linspace(start, end, num_bins)
plt.hist(data1, bins=bins, alpha=0.5, label='Dataset 1')
plt.hist(data2, bins=bins, alpha=0.5, label='Dataset 2')


Histogram normalization

When comparing datasets of different sizes, histograms should be normalized to compare their shapes:

plt.hist(data1, bins=bins, alpha=0.5, density=True, label='Dataset 1')
plt.hist(data2, bins=bins, alpha=0.5, density=True, label='Dataset 2')


Normalization makes the area under the histogram curve sum to 1, ensuring comparability.

Dealing with different scales

For histograms with different scales, twinx() can be used to create a secondary y-axis:

fig, ax1 = plt.subplots()
ax2 = ax1.twinx()
ax1.hist(data1, alpha=0.5, label='Dataset 1')
ax2.hist(data2, alpha=0.5, label='Dataset 2', color='red')


Using colors and labels

Distinguishing your datasets with color and label brings clarity and meaning:

plt.hist(data1, bins=bins, color='skyblue', alpha=0.7, label='Data group 1')
plt.hist(data2, bins=bins, color='salmon', alpha=0.7, label='Data group 2')


Labels are key when you're dealing with overlaid histograms. Always pair them with the plt.legend() function.

Advanced techniques and troubleshooting

Preventing data overlap

Ensure no histogram hides the other:

• Shift the bin edges.
• Use transparent colors.
• Adjust the zorder parameter.

Using weights in histograms

To balance differently-sized samples, weights can be helpful:

# Assume weight1 and weight2 are the weights for each data point
plt.hist(data1, weights=weight1, bins=bins, alpha=0.6, label='Weighted Dataset 1')
plt.hist(data2, weights=weight2, bins=bins, alpha=0.6, label='Weighted Dataset 2')


Dynamic data for examples

For illustration, random.gauss() can be used to generate data:

import random

# Generate random data
data_demo = [random.gauss(mu, sigma) for _ in range(1000)]


Clearing axis for new plots

Clear the axis to avoid confusion with old plots:

plt.cla()  # Clears the current axes like an over-enthusiastic housemaid
plt.clf()  # Clears the entire figure like a spy wiping their existence
