Welcome

A practical, intuition-first statistics course for data scientists who already know Python and Pandas — built around reasoning under uncertainty, not memorizing formulas.

Welcome to Statistics for Data Science with Python. You already know how to load a CSV, filter a DataFrame, group it, and chart it. You can answer "what does the data say?" This course teaches the harder, more valuable question: "how much should I trust what the data says?"

That second question is where statistics lives. Two teams can run the exact same query on the exact same table and walk away with opposite conclusions — one ships a feature, the other kills it — because one of them mistook a random wiggle for a real signal. Statistics is the discipline that keeps you from fooling yourself.

What we assume you already know

You can write basic Python, use Pandas (read_csv, filtering, groupby, merge), and make a chart. You do not need any prior statistics, probability, or heavy math. We build all of that from intuition.

This is a reasoning course, not a formula course

Most statistics material is a wall of symbols: estimators, theorems, and derivations you memorize for an exam and forget by Friday. This course is the opposite. For every idea we ask:

What problem does it solve? Why was it invented?
When should you use it — and when should you not?
What do people get wrong about it?
Where does it show up in real data-science work?

You will still see formulas, but only when a formula makes the intuition clearer, never as something to memorize. The computer does the arithmetic; your job is to know which question you're asking and whether the answer makes sense.

The mantra for this whole course

Data describes the sample you happened to collect. Statistics is how you reason about the population you actually care about. Almost every concept here is a tool for crossing that gap honestly.

What you'll be able to do by the end

By the time you finish, you'll be able to:

Reason about uncertainty, variation, and randomness instead of treating every number as exact.
Pick the right summary statistics and know when a mean lies to you.
Use probability and probability distributions to model how data behaves.
Understand sampling — why a sample of 1,000 can speak for millions, and how it goes wrong.
Build and interpret confidence intervals and the bootstrap.
Run and correctly interpret hypothesis tests with scipy.stats.
Explain what a p-value really is (and the four things it is not).
Distinguish statistical significance from practical significance using effect sizes.
Spot the classic statistical fallacies that sink real analyses.
Run a credible A/B test and a disciplined exploratory statistical analysis.

How the course is organized

We move from describing data, to modeling uncertainty, to making inferences and testing claims, and finally to applying it all.

Each section builds on the one before it. The middle of the course — sampling distributions, the central limit theorem, confidence intervals, and p-values — is the conceptual heart. If those four ideas click, everything else is application.

The tools we use

Every code block on every page is runnable. Edit it, click Run, and the output appears underneath. We lean on four libraries you'll use constantly as a working data scientist:

Library	What we use it for
NumPy	Generating data, vectorized math, random sampling
pandas	Real datasets, grouping, summaries
scipy.stats	Distributions, hypothesis tests, confidence intervals
Plotly Express	Visualizing distributions, sampling, and results

Here's a taste. This snippet simulates an A/B test where the two groups are truly identical, then asks scipy.stats whether it can tell them apart. Run it a few times in your head — sometimes the "difference" looks convincing even though there's nothing there.

By the end of the course, every line of that snippet — normal, ttest_ind, the p-value, and especially how to interpret it — will be second nature.

How the interactive widgets work

You'll meet three kinds of interactive elements:

Code blocks — editable, runnable Python. Change anything, re-run, experiment.
Challenge cards — small problems with hidden tests. Write a solution, click Submit, and see which tests pass. These are where the learning sticks, so do them.
Multiple-choice questions — quick conceptual checks with an explanation for every option.

Each code block runs on its own

Variables you define in one code block are not shared with the next one, even on the same page. Every block starts fresh, so each example is self-contained.

A note on honesty

Statistics has a reputation for being a tool people use to lie. The truth is the opposite: statistics is the toolkit for not lying — to others, and especially to yourself. The whole field is a set of disciplined habits for staying humble about what a pile of numbers can and cannot prove.

Let's begin with the most important question of all: if we already have the data, why isn't the data enough?