python-scipyHow do I use Python SciPy to bootstrap a dataset?

Bootstrapping is a powerful technique for estimating statistics on a dataset by resampling with replacement. SciPy provides a convenient function scipy.stats.bootstrap to do this.

Example code

import numpy as np
from scipy.stats import bootstrap

data = np.array([1, 2, 3, 4, 5, 6, 7, 8])
bootstrap_means = bootstrap(data, 1000, np.mean)

The above code will generate 1000 bootstrap samples of the dataset and calculate the mean of each sample. The output of the code will be a numpy array of 1000 means.

Code explanation

1. import numpy as np: Imports the numpy library as np.
2. from scipy.stats import bootstrap: Imports the bootstrap function from the SciPy stats module.
3. data = np.array([1, 2, 3, 4, 5, 6, 7, 8]): Creates an array of the data to be bootstrapped.
4. bootstrap_means = bootstrap(data, 1000, np.mean): Calls the bootstrap function with the data, the number of bootstrap samples to generate, and the function to use to calculate the statistic (in this case, the mean).

Helpful links

• scipy.stats.bootstrap
• Bootstrapping

Edit this code on GitHub