Sampling without replacement in Python

Using the NumPy package's random.choice() function

In statistics and probability, we often randomly draw items out of a group. A common example is choosing names out of a hat to determine the winners of a prize. You can use the NumPy package in Python to do this.

Here is an example of such sampling without replacement. It uses the “random.choice()” function from the NumPy package. Note the "replace=False" option. The following code chooses 3 names out of 5:

import numpy as np

names = ['Anne', 'Bill', 'Cindy', 'Evan', 'Fiona']
sample_names = np.random.choice(names, 3, replace=False)

display(sample_names)

Of course, if you want to sample with replacement, you simply use the option "replace=True" instead.

There are practical applications of this function in statistics, data analysis, and data science. Here are 3 examples.

A/B Testing

Suppose that you want to test two different versions of a website (A and B) to see which one leads to higher conversion rates. You can randomly assign incoming users to either website A or website B using np.random.choice(). The option replace=False is suitable to the assignment process, because a user can only experience one version at a time. This random sampling ensures that the 2 groups are as similar as possible, minimizing bias and allowing for a fair comparison.

Survey Research:

If you conduct a survey but cannot interview every person in a large population, then you can use np.random.choice() to select a random and representative sample of individuals from the population to participate in the survey. The option replace=False ensures that the same individual isn't accidentally contacted multiple times.

Quality Control

If you work at a manufacturing plant and want to assess the quality of your products, you can use np.random.choice() to choose a random sample of products for inspection. Then, each product will have an equal chance of being selected, and no product will be inspected twice from the same sampling effort.