A course that finishes
Ten modules in the order the work happens — from what a row means to what wakes you up at 3am. Every lesson ends with an exercise you can run against a dataset on this site.
Datanemics
A crash course in the work data teams actually do.
No fluff, no paywall, no prerequisites beyond a terminal. Ten modules in the order the job happens, 24 patterns you can look up mid-task, and 7 tools that run in your tab.
- Lessons
- 0
- Patterns
- 0
- Datasets
- 0
- Rows
- 0
- Tools
- 0
01What is here
Patterns you can look up
Named problem shapes — fan-out joins, censored cohorts, leakage guards. Each names the version that looks right and is not, then the one that holds.
Datasets with the mess left in
Each one is built around a specific analytical trap: informative missingness, censored durations, a paradox that reverses on disaggregation, five date formats in one column.
Tools that run in your tab
A SQL engine, a chart builder, a regression lab, a distribution sandbox. All client-side, working on the datasets above or on a CSV you drop in.
02The syllabus
The syllabus
Built backwards from the job. You get the data, you find out it is wrong, you model it, you measure something, you tell someone, and then you keep it running. Each module stands alone — but the order is the point.
014 lessons
Foundations
What the job actually is, what a row means, and how to turn a vague request into a question with an answer.
024 lessons
Getting the data in
Ingestion, batch versus streaming, idempotency, and what happens when yesterday arrives twice.
034 lessons
Modelling the warehouse
Facts, dimensions, grain, slowly changing dimensions, and why columnar storage decides your query cost.
044 lessons
SQL that survives production
Joins that do not fan out, window functions, the NULL rules that catch everyone, and how to read a query plan.
053 lessons
Data quality and trust
Testing data like code, the anatomy of a silent failure, and how to find out before your stakeholder does.
064 lessons
Statistics you will actually use
Distributions, uncertainty, sampling, and the small number of tests that cover most real questions.
073 lessons
Experiments and causal claims
Designing a test that can succeed, reading a result honestly, and what to do when you cannot randomise.
084 lessons
Modelling, honestly
Baselines, leakage, validation that matches deployment, and choosing a metric that means something.
093 lessons
Showing the result
Choosing a form, using colour as an encoding rather than decoration, and writing the sentence the chart is making.
103 lessons
Shipping it and keeping it alive
Orchestration, monitoring, cost, and the on-call reality of owning something that runs every night.
03The toolkit
The toolkit
Every tool runs entirely in your browser against the datasets on this site — or against a CSV you drop in, which never leaves your machine.
Chart Builder
Point it at any dataset, choose an x, a y, and a grouping, and get a chart that follows the rules — one axis, direct labels, a real legend, and a table view underneath.
Data Explorer
A fast table with per-column profiling: type inference, missing counts, distinct values, quantiles, and a sparkline distribution for every numeric column.
SQL Playground
Write real SELECT queries against any dataset here and get results instantly — joins, aggregates, GROUP BY, HAVING and ORDER BY, with plain errors for the things it does not support.
Distribution Lab
Sample from seven distributions, watch the central limit theorem happen at whatever sample size you choose, and see exactly when the mean stops being a useful summary.
Correlation Explorer
A diverging heatmap of every numeric pair in a dataset, with a click-through scatter so you can check whether the number is telling the truth.
Regression Lab
Fit an ordinary least squares line to any pair of columns and get the diagnostics that decide whether the fit means anything: residual plot, R², standard errors, and leverage.
Experiment Calculator
Sample size before you launch, and a confidence interval after you stop. Both directions, with the assumptions written on the screen rather than buried in a footnote.
04The dataset library
The dataset library
14 synthetic datasets, 72,860 rows. Each page names exactly what is buried in the data and what to try against it.
Ride-hail trips
8,000 rows10 cols
Retail orders
7,000 rows12 cols
SaaS subscriptions
2,500 rows10 cols
Factory sensor telemetry
8,000 rows8 cols
Flight delays
8,000 rows10 cols
Randomised trial results
900 rows11 cols
City air quality
5,480 rows8 cols
Data job postings
3,500 rows11 cols
05Who writes it
Who writes it
Nine practice areas, each owning the material closest to its day job. Every lesson and post carries the discipline it came from.
DE
Data engineering
Ingestion, orchestration, and the unglamorous parts that keep everything else standing.
AE
Analytics engineering
Turning warehouses full of raw tables into models an analyst can trust without asking anyone.
EX
Experimentation
Designing tests that can succeed, and reading the results honestly.
ST
Statistics
Uncertainty, sampling, and the difference between a result and a finding.
ML
Machine learning
Getting models out of notebooks and into places where they can do damage responsibly.
DQ
Data quality
Tests, contracts, reconciliation, and finding out before the CFO does.
06Recent writing
Recent writing
Long-form articles on how the tools work, and shorter field notes from practice.
1 min read
The overall average hides 8 different numbers
Real group means for one column across 8 segments, and what the pooled average conceals.
1 min read
A essentially none correlation, and what it is not
A measured correlation of -0.00 between two columns, what its square says, and the diagnostic that costs one line.
1 min read
What 3,500 job postings say about city
One column of a postings dataset, and what the shape of it means for anyone reading listings.