Data Science and AI
Statistics, Python and machine learning kept to problems small enough to finish and honest enough to trust.
- Courses
- 18
- Hours
- 26
- Modules
- 5
The first three courses contain no machine learning at all. That ordering is deliberate: most analysts skip to the specialist tools while still getting variation wrong.
Assessed on a notebook: a question, the analysis, the uncertainty, and the conclusion you are willing to defend.
The sequence
Modules unlock in order. You finish one before the next opens, which is the point of a path rather than a playlist.
-
01
Variation first
How much a number moves on its own, before you interpret it moving.
-
02
Uncertainty as a range
Bootstrapping, and why a point estimate is half an answer.
-
03
Base rates
Why a ninety per cent accurate screen mostly flags the innocent.
-
04
Python for analysis
The small part of the ecosystem that does most of the work.
-
05
Machine learning, honestly
Applied to problems you can finish, with what it cannot do stated.
What you can do afterwards
The assessment checks these directly, against a rubric published before you start.
Before you start Comfortable with a scripting language and unafraid of arithmetic.
- Plot the last twenty periods before interpreting this one.
- Report an estimate with its range.
- Say out loud what your model cannot do.