Free chapters
One complete chapter from each course, exactly as paying students read it. No sign-up, no email required. If the writing does not work for you, the course will not either.
- 1
Data Analysis Foundations
What a good analysis looks like
Most analysis at work fails before any number is computed. Someone is handed an export, opens it, and starts making charts. Two hours later there are eleven charts and no answer. The fix is a habit, not a tool: write the…
Read the chapter - 2
SQL for Analysts
Tables, rows, keys
A database is a set of tables, and a table is nothing more than a spreadsheet with rules. Every row is one thing: one order, one customer, one payment. Every column is one fact about that thing. The rules are what make S…
Read the chapter - 3
Statistics for Business Decisions
Variation is normal
Every metric you track moves every day, and most of that movement means nothing. Conversion was 3.1% on Monday and 2.8% on Tuesday. Someone asks what went wrong. Usually the honest answer is 'nothing; that is what conver…
Read the chapter - 4
Python for Data Analysis
DataFrames
A DataFrame is a table: rows, columns, and an index that labels the rows. If you have used a spreadsheet or SQL, nothing here is new except the vocabulary. df.head() shows the first rows. df.shape tells you how many rows…
Read the chapter - 5
Dashboards & Data Storytelling
Dashboards versus analyses
A dashboard is for a question that gets asked every week and has the same shape every time: how are sales tracking against target, what is the refund rate by product, how many support tickets are open. An analysis is for…
Read the chapter - 6
Fraud Analytics
How fraud shows up in data
Fraud is a story about incentives, but it leaves a trail in tables. A stolen card is used fast, on new devices, for goods that can be resold, before the real owner notices. An account takeover changes the phone number or…
Read the chapter - 7
Machine Learning for Fraud Detection
Labels
A fraud model learns from examples of fraud, and the examples are worse than they look. A chargeback arrives 30 to 120 days after the transaction, so the most recent months of data have almost no fraud labels, not becaus…
Read the chapter - 8
Applied AI for Analysts
Where a model fits
Language models are good at tasks where the input is messy language and the output is language or a simple structure: pull the company name and address out of this paragraph, decide which of these eight categories this c…
Read the chapter