In the Age of AI — FRONTLINE Documentary
2 hours
FRONTLINE's documentary on artificial intelligence, its economics and who carries the cost.
Data science here means the whole path from a raw table to something somebody can act on: cleaning and shaping data, the statistics that stop you fooling yourself, and the charts that carry a finding to people who were not in the room. Data visualisation has a page of its own beneath this one, covering the individual tools in depth.
Everything listed under data science is free to start, and most of it is pitched at beginners, with a smaller set of intermediate and advanced courses for people who already work with data and want a specific gap filled. The courses come from YouTube, where the teaching is done in public and you can judge an instructor before spending an evening with them.
Each course is compared on the same terms: level, length, price, certificate and the source it comes from. Filter by level to skip the introductions if you have already done them.
120 courses
2 hours
FRONTLINE's documentary on artificial intelligence, its economics and who carries the cost.
18 minutes
Continuous integration applied to machine learning, in under twenty minutes.
5 hours
Five hours on MLOps: getting a model out of a notebook and keeping it running.
12 hours
Twelve hours of Python, taught by building rather than by explaining.
12 hours
Twelve hours of Python for beginners, taken slowly enough to follow along.
4 hours
Four and a half hours of Python from nothing, the course most people start with.
55 minutes
Confusion matrices, precision, recall and ROC curves, worked through in scikit-learn.
45 minutes
Forty-five minutes of scikit-learn, from fitting a model to scoring it.
32 minutes
A university lecture on evaluating a model: the measures, and what each one hides.
13 minutes
Thirteen minutes on how a model is judged, and why accuracy alone is rarely the answer.
55 minutes
Building a working e-commerce data pipeline, start to finish, with Snowflake and dbt.
3 hours
Nearly three hours on time series: the components, the models, and the forecasts built from them.
2 hours
An hour and a half of forecasting in Python, for somebody who has not done it before.
2 hours
An hour and a half on association rules: finding what goes with what in a transaction log.
2 hours
Two and a half hours on decision trees — how one is built, and how it is read.
4 hours
Three and a half hours on clustering: k-means, hierarchical methods, and choosing between them.
10 hours
Ten hours of data mining: the techniques, the algorithms and what each is for.
13 minutes
What machine learning is, explained without the mathematics, in thirteen minutes.
22 minutes
How categorical columns become numbers a model can use, and what goes wrong when they do not.
1 hour
Natural Language Processing (NLP)
An hour on the step every NLP pipeline starts with: turning text into something a model can read.
New courses, a couple of emails a month. One click unsubscribes.