0 of 6 done
Start hereDescriptive Statistics

Describing data you have, and reasoning about data you do not. Written to the course standard: prerequisites, observable outcomes, worked derivations, a runnable lab whose output was generated from an actual run, practice with hidden answers, and a demonstrable finish line.

This umbrella merges both curricula — SS1–SS3 statistics and probability from the NERDC outline, and the university probability and mathematical statistics course from the university track — so each subject appears once, secondary treatment first.

The one idea: every summary is a compression, and compression loses information. A mean hides a bimodal split; a -value hides an effect size; a confidence interval hides that the describes the procedure rather than the interval. This course teaches the tools and, in the same breath, exactly what each one throws away.

Reading order

Strictly sequential. Inference needs distributions, which need probability, which needs the descriptive vocabulary.

  1. Descriptive Statistics — [Beginner] — mean, median, mode, grouped data, and Anscombe’s quartet: four datasets agreeing on every statistic and looking nothing alike
  2. Dispersion and Cumulative Frequency — [Beginner] — range, IQR, variance, ogives, outliers, and why the variance formula divides by , demonstrated rather than asserted
  3. Probability — [Beginner → Intermediate] — sample spaces, the rules, conditional probability, Bayes, and why a 99%-accurate test is wrong 91% of the time it fires
  4. Random Variables and Distributions — [Intermediate] — expectation, variance, and the five distributions worth knowing, each tied to the mechanism that produces it
  5. Inference and Estimation — [Advanced] — the Central Limit Theorem, standard error, confidence intervals, hypothesis tests, and what a -value is not
  6. Stochastic Processes — [Advanced] — Markov chains, random walks, queues, and why the last 10% of capacity is unusable

Where the two curricula meet

LessonSecondary sourceUniversity source
Descriptive statisticsSS1 T3 — frequency tables, charts, mean/median/modeassumed
Dispersion, cumulative frequencySS2 T3 — ogives, range, variance, standard deviationassumed
ProbabilitySS2 T3 basic; SS3 T2 mutually exclusive, independent, tree diagramsY3S2 probability spaces
Distributions—Y3S2 — random variables, Normal, Binomial, Poisson
Inference—Y3S2 — central limit theorem, parameter estimation
Stochastic processes—Y4S2 — Markov chains, Poisson processes

What is verified

Every lab was executed and its expected output generated from that run. Several exist to test a claim rather than illustrate one:

  • Anscombe’s quartet — seven statistics agreeing across four datasets to within , and text scatter plots showing a line, a parabola, an outlier and a vertical stack.
  • The correction — 200,000 samples of size 5: dividing by lands at exactly of the true variance, which is .
  • The base rate fallacy — counted out over 100,000 people: 99 true positives against 999 false ones.
  • Monty Hall with a control — a knowing host gives / ; a host opening at random gives no advantage at all, isolating knowledge as the mechanism.
  • Poisson as a binomial limit — the gap shrinking by a factor of ten for every tenfold increase in .
  • Confidence interval coverage — 10,000 intervals per sample size, showing -intervals cover only at and still fall marginally short at .
  • -values under a true null — uniform to the decile, with falling below .
  • M/M/1 queueing — a traffic rise multiplying the queue eleven-fold.

How to study this

  1. Attempt the prediction before reading on. Every lesson stops once and asks you to commit. The answers sit at the end of Check your understanding, never beside the question.
  2. Trust the count over the formula. Bayes is easier as a table of 100,000 people than as an equation, and the lessons do it that way first.
  3. Simulate to check, derive to claim. Every simulation here carries error; it confirms arithmetic, it does not establish results.
  4. Watch what each tool discards. That is the thread through all six lessons, and the reason the labs are built the way they are.
  • mathematics/ — the parent course and the full topic map
  • Combinatorics — the counting probability is built on
  • Eigenvalues — the steady state of lesson 6, algebraically
  • ai-ml/ — where distributions, inference and the CLT are used daily
  • Course standard — the teaching shape these lessons follow
↑ Back to Mathematics