Statistical Functions in NumPy

Why statistics in NumPy?

Quick descriptive stats help you:

Understand distributions
Detect outliers
Summarize data before modeling/visualization

These aggregation methods (also called reductions) are available both as array methods (arr.mean()arr.mean()) and as top-level NumPy functions (np.mean(arr)np.mean(arr)) — they do the same thing.

Sample data

data

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])

data

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])

Mean, median

mean-median

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.mean(data))
print(np.median(data))

mean-median

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.mean(data))
print(np.median(data))

Min, max, range

min-max

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.min(data))
print(np.max(data))
print(np.ptp(data))  # peak-to-peak = max - min

min-max

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.min(data))
print(np.max(data))
print(np.ptp(data))  # peak-to-peak = max - min

Variance and standard deviation

var-std

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.var(data))
print(np.std(data))

var-std

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.var(data))
print(np.std(data))

Percentiles / quantiles

percentile

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.percentile(data, 25))
print(np.percentile(data, 50))
print(np.percentile(data, 75))

percentile

import numpy as np
 
data = np.array([12, 15, 14, 10, 25, 19, 18])
 
print(np.percentile(data, 25))
print(np.percentile(data, 50))
print(np.percentile(data, 75))

diagram From raw data to a summary mermaid

  flowchart LR
  A["Raw array"] --> B["Center: mean, median"]
  A --> C["Spread: var, std, ptp"]
  A --> D["Position: percentile, quantile"]
  A --> E["Relationship (2 arrays):
cov, corrcoef"]

A handful of NumPy reductions turn a raw array into the descriptive statistics analysts reach for first.

Working across axes (2D)

axes

import numpy as np
 
mat = np.array([
    [1, 2, 3],
    [4, 5, 6]
])
 
print(np.mean(mat, axis=0))  # per column
print(np.mean(mat, axis=1))  # per row

axes

import numpy as np
 
mat = np.array([
    [1, 2, 3],
    [4, 5, 6]
])
 
print(np.mean(mat, axis=0))  # per column
print(np.mean(mat, axis=1))  # per row

Correlation and covariance

corr-cov

import numpy as np
 
x = np.array([1, 2, 3, 4, 5])
y = np.array([2, 4, 6, 8, 10])
 
cov = np.cov(x, y)
print("cov:\n", cov)
 
corr = np.corrcoef(x, y)
print("corr:\n", corr)

corr-cov

import numpy as np
 
x = np.array([1, 2, 3, 4, 5])
y = np.array([2, 4, 6, 8, 10])
 
cov = np.cov(x, y)
print("cov:\n", cov)
 
corr = np.corrcoef(x, y)
print("corr:\n", corr)

Continue to: Saving and Loading NumPy Data to persist arrays efficiently.

🧪 Try It Yourself

Exercise 1 – Mean and Median

Exercise 2 – Standard Deviation

Exercise 3 – Median via the 50th Percentile

If this helped you, consider buying me a coffee ☕

Buy me a coffee