Teach machine learning as a disciplined workflow rather than a collection of algorithms. This Unit 05 pack transforms Chapter 5 of AI & Machine Learning: Foundations & Data – Volume 1 into four structured, classroom-ready lessons covering ML vs rule-based systems, supervised, unsupervised and reinforcement learning, train/validation/test discipline, data leakage, underfitting and overfitting, first models, preprocessing, pipelines, cross-validation and evaluation metrics.

Students learn to define a target, identify available evidence, protect the test set, recognise high-bias and high-variance behaviour, distinguish regression from classification, choose preprocessing steps and select metrics that match the cost of errors. The unit also introduces a first scikit-learn model through the create → fit → predict → evaluate pattern using LinearRegression, with MAE, RMSE and R² used to support cautious model claims.

Designed as the classroom teaching companion to AI & Machine Learning: Foundations & Data – Volume 1, the package includes a Teacher Guide, four-lesson plan, Teacher Handbook, Student Summary Notes, printable Worksheet, full Answer Key, Activity Cards, Exit Tickets, an editable 34-slide presentation deck and a Read Me First guide. Optional code-reading tasks allow the unit to work with or without student devices.

The no-prep structure supports direct teaching, group work, discussion, assessment and independent revision. The central message is simple: a model should not be trusted because it sounds advanced; its target, evidence, split, features, metric and limitations must make sense.

By the end of this unit, students will be able to:

• Distinguish machine learning from rule-based programming
• Identify supervised, unsupervised and reinforcement learning
• Define a prediction target and identify useful evidence
• Explain train, validation and test roles
• Identify data leakage
• Recognise underfitting/high bias and overfitting/high variance
• Distinguish regression from classification
• Read a LinearRegression create → fit → predict → evaluate workflow
• Explain scaling, encoding, imputation and pipelines
• Explain the purpose of cross-validation
• Compare key regression and classification metrics
• Choose metrics based on prediction type and error cost
• Write a responsible model claim with a limitation

Lesson 1 – From Rules to Learning
Compare hand-written rules with learned patterns and classify examples by learning paradigm.

Lesson 2 – The ML Workflow and Honest Splits
Follow the workflow, protect train/validation/test roles and repair leakage risks.

Lesson 3 – First Models and Model Failure
Compare regression and classification, interpret bias-variance behaviour and read a first Linear Regression workflow.

Lesson 4 – Features, Pipelines and Evaluation
Choose preprocessing steps and metrics, use pipeline reasoning and connect cross-validation to trustworthy evaluation.