SciPy stats intro

The scipy.stats module provides probability distributions, descriptive statistics, and hypothesis tests on NumPy arrays—essential for data science after Pandas cleaning.

What scipy.stats offers

• rv_continuous / rv_discrete — distribution objects with pdf, cdf, ppf, rvs
• describe, tstd, skew, kurtosis — descriptive summaries
• ttest_ind, mannwhitneyu, chi2_contingency — hypothesis tests
• pearsonr, spearmanr — correlation coefficients

Distribution object pattern

import numpy as np
from scipy import stats

dist = stats.norm(loc=0, scale=1)
print('pdf(0):', dist.pdf(0))
print('cdf(1.96):', dist.cdf(1.96))
print('sample:', dist.rvs(size=5, random_state=0))

Arrays in, results out

Pass 1D NumPy arrays of observations. Most tests return a result object with statistic and pvalue attributes you log or report in notebooks.

Important interview questions and answers

1. Q: stats vs np.mean?
   A: NumPy aggregates; stats adds distributions, tests, and standardized inference helpers.
2. Q: rvs meaning?
   A: Random variates—draw samples from the distribution.

Self-check

1. What four capabilities does scipy.stats provide?
2. Name three methods on a distribution object.

Tip: Distribution objects share pdf/cdf/ppf/rvs—learn one family, know them all.