Course status. Module 1 is open for review and Modules 2 to 8 are in preparation. There are no accounts, no saved progress and no certificate issuing yet: everything you answer stays in your browser tab and is not stored.
Open Module 1
Course Passport
| Item | Entry |
|---|---|
| Course title | Learning from Data: Principles of Predictive Models |
| Track | University Track, one of four (Foundation, University, Professional and Executive) |
| Intended learner | Undergraduates and early-career graduates, and any adult with a quantitative background who wants to understand machine learning without heavy mathematics. |
| Length | 8 modules, one per week, about 6 study hours each: about 48 hours in total |
| Format | Interactive visual lessons (animated infographics, tables, graphs). No video lectures or podcasts. |
| Prerequisites | Comfort with percentages, averages and reading a table or graph. No programming is required. |
| Language and tone | Formal academic English. Every technical term is defined at first use. |
| Assessment | Practice sets 10%, module checkpoints 40%, labs and cases 15%, capstone 35% |
| Pass rule | 70% overall, and the capstone rubric passed separately. Checkpoints unlock the next module at 80%, with unlimited retakes. |
| Credential | Certificate of Completion. It states that it is not an accredited qualification. |
Why this course exists
Predictive models now shape decisions from credit to medicine, yet many users cannot say how a model learns or how to judge its claims. This course teaches the principles that apply to any model: examples, features and labels, error, baselines, training and test data, and over-fitting. It relies on small tables you can check by hand rather than on software. For background, see the encyclopedia entries on machine learning, inference and regression analysis.
Learning outcomes
By the end of the course, a learner can:
- O1. Describe supervised learning in terms of examples, features, labels and predictions.
- O2. Measure a model’s error and compare it with a simple baseline.
- O3. Fit and interpret a simple line or rule on a small table.
- O4. Explain training, validation and test data, and over-fitting.
- O5. Evaluate a classifier with accuracy, precision and recall.
- O6. Critique a model’s claims, including bias, leakage and limits.
Course map
Eight modules move from examples and features, through error and baselines, to fitting, classification and generalisation, and end with the learner’s own small model, evaluated honestly.
| Module | Big idea | Hours | Outcomes | Status |
|---|---|---|---|---|
| M1. From Data to Prediction: Examples, Features and Labels | A model is a rule that maps known features to a prediction. | 6 | O1, O2 | Open for review |
| M2. Measuring Error: Loss, Accuracy and Baselines | You cannot judge a model without a number and a comparison. | 6 | O2 | In preparation |
| M3. Fitting a Line: Regression | The best line minimises error on the data it sees. | 6 | O3 | In preparation |
| M4. Classification and Decision Boundaries | A classifier separates cases into classes. | 6 | O3 | In preparation |
| M5. Generalisation: Training, Validation and Test Data | A model must work on data it has not seen. | 6 | O4 | In preparation |
| M6. Evaluating Classifiers: Confusion Matrix, Precision and Recall | Accuracy alone can hide serious mistakes. | 6 | O5 | In preparation |
| M7. Models in Practice: Bias, Leakage and Limits | A model reflects the data and the choices behind it. | 6 | O6 | In preparation |
| M8. Your Small Model: Build, Test and Report | A small model, evaluated and reported honestly. | 6 | O3, O4, O6 | In preparation |
Capstone thread (proposed). The capstone is a small predictive model on a table of the learner’s choice: the question and baseline are set in Module 1 and Module 2, the model is fitted and tested in Modules 3 to 5, and the evaluation with limits is reported in Module 8.
Limits of this course
The course teaches the principles of predictive models as a field of study. It uses small invented tables, does not require programming, and does not prepare learners to deploy systems that affect people. Results on tiny tables are used to teach ideas, not to establish facts.
Our standard
Numbers are checked and sources are named by title and publisher. Limits are stated. The course awards no degree and no academic credit. Completing it will earn a Certificate of Completion only, which states plainly that it is not an accredited qualification. Certificates are not issued yet.
More courses: see all Specialty Digest Online Courses or the University Track.