PythonPython · 4 weeks
Python, gently
Variables, lists, loops and functions, taught to people who have never written a line of code. By the last week you load a spreadsheet, tidy it and answer a real question with it.
The full list
24 courses, from your first line of code to work you can publish.
Every course is taught live to 10 people or fewer and costs $9.50 to $186.00. Asking for a place only starts a conversation; you never pay through this site.
Every course
Showing 24 of 24.
Beginner$9.50
Pythonpandas · 3 weeks
Tidy tables with pandas
Reading, filtering, grouping and reshaping data frames. You learn what tidy data means, and why it makes every later step easier.
Pythonuv · 4 weeks
Python other people can run
Virtual environments, packaging, tests and a project layout a colleague can open next year. For anyone whose analysis currently works on one machine only.
PythonPolars · 4 weeks
Python for larger datasets
Vectorising with NumPy, switching to Polars, and profiling before guessing. You take a script that runs overnight and make it finish before lunch.
Data & SQLSQL · 3 weeks
SQL for the curious
SELECT, WHERE, joins and GROUP BY against a database of real public records. A friendly first look at asking a database a question directly.
Data & SQLpandas · 3 weeks
Cleaning messy data
Dates in five formats, names spelt three ways, and duplicates hiding in plain sight. You write down each cleaning step so it can be checked and repeated.
Data & SQLDuckDB · 3 weeks
DuckDB on your own laptop
Analytical SQL over Parquet and CSV files without a server. Window functions and fast queries on tables bigger than a spreadsheet can open.
Data & SQLdbt · 5 weeks
A small, sturdy data pipeline
Ingest, transform and test data on a schedule with dbt and a little Python. Aimed at the person who refreshes the same report by hand every Monday.
StatisticsStatistics · 4 weeks
Statistics without the dread
Averages, spread, sampling and confidence intervals, explained with pictures and plain words before any formula. For people who were put off statistics at school.
StatisticsR · 5 weeks
Regression, carefully
Linear and logistic models, what the coefficients actually say, and the checks to make before you believe them. Taught in R, with Python notes alongside.
Statisticsstatsmodels · 4 weeks
Experiments and A/B tests
Randomisation, power, p-values and the experiment that stopped too early. You design a test, size it properly and write up an honest result.
StatisticsPyMC · 5 weeks
Bayesian thinking in practice
Priors, posteriors and hierarchical models fitted with PyMC. For analysts who want to say how sure they are, not only whether something is significant.
Visualisationmatplotlib · 2 weeks
Charts that tell the truth
Choosing the right chart, starting axes honestly and labelling so nobody has to guess. Two short weeks that improve every figure you make afterwards.
Visualisationggplot2 · 3 weeks
Figures with ggplot2
The grammar of graphics, layer by layer, until a publication figure takes minutes instead of an afternoon. Taught in R for researchers who submit papers.
VisualisationObservable · 4 weeks
Interactive figures for the web
Tooltips, filters and small linked views with Observable Plot. You publish a figure that readers can explore without installing anything.
VisualisationDashboards · 3 weeks
A dashboard people actually read
Deciding what belongs on the page, what to leave off, and how to keep it accurate as the data changes. For people who own a dashboard that nobody opens.
Machine learningPython · 4 weeks
Machine learning from first principles
What a model learns, why it can be wrong, and how to tell. Decision trees and nearest neighbours built by hand before any library does it for you.
Machine learningscikit-learn · 5 weeks
scikit-learn, step by step
Pipelines, feature preparation, model choice and tuning on a realistic dataset. You finish with a model you can explain to the person relying on it.
Machine learningscikit-learn · 3 weeks
Evaluating models honestly
Baselines, leakage, cross-validation and metrics that match the real cost of a mistake. A short course in not fooling yourself.
Machine learningLLMs · 5 weeks
Working carefully with language models
Embeddings, retrieval, structured output and evaluation sets for research text. You build a small tool, then measure how often it is right before trusting it.
Research practiceGit · 2 weeks
Git for researchers
Commits, branches and GitHub, explained with your own analysis rather than someone else's software. Never again a folder of files called final_v3.
Research practiceJupyter · 3 weeks
From notebook to reproducible project
Turning a long notebook into a project that rebuilds every number from the raw data with one command. Calm, practical and useful in any field.
Research practiceQuarto · 3 weeks
Writing up an analysis with Quarto
Text, code and figures in one document that renders to a paper, a report or a website. Your numbers can no longer drift away from your prose.
Research practiceOpen research · 3 weeks
Sharing data and code responsibly
Licences, anonymisation, data archives and documentation that another researcher can follow. For people preparing work for publication or a funder's review.
A note from the tutors
Not sure where to begin? Write to us.
Tell us a little about your work and the thing you keep meaning to learn. We'll suggest a course, or tell you honestly that you don't need one.
A person reads every message sent to [email protected], and we always reply.
With best wishes,
the tutors at Collingwood