Use ddof=1 for Sample Dispersion Measures with NumPy

Zettel#144

zettelkasten
Published

2024-03-01

By default, np.var() and np.std() calculate the population (biased) variance and standard deviation, whereas statistics.variance() and statistics.stdev() calculate the sample (unbiased) variance and standard deviation by default.

However, you can calculate the unbiased variance and standard deviation with NumPy by setting the ddof parameter to 1.

import numpy as np
import statistics

# Using NumPy
population_variance = np.var(data)
sample_variance = np.var(data, ddof=1)

# Using statistics
sample_variance = statistics.variance(data)
sample_standard_deviation = statistics.stdev(data)