Predictive Modelling (Supervised Learning)

All models are wrong, but some are useful. — George Box

Introduction

Predictive modelling uses labelled data to learn patterns and make predictions. This topic covers the core supervised learning algorithms.

What You Will Learn

• Implement and compare linear and logistic regression models
• Train decision tree and ensemble models
• Apply SVM for classification tasks
• Select appropriate algorithms based on problem characteristics

Assessment Connection

Section A — you must "compare two ML models/approaches" with clear justification for selection.

Content

TutorialsHow-to GuidesReferenceExplanation

• Linear Regression
• Logistic Regression
• Decision Trees
• Random Forests
• Support Vector Machines
• Gradient Boosting
• Neural Networks
• Model Comparison

• Build a Baseline Model
• Handle Imbalanced Data
• Multi-Class Classification
• Prevent Overfitting
• Run Regression Diagnostics
• Save and Load Models

• Algorithm Selection Flowchart
• Loss Functions Reference
• Sklearn Classifiers API
• Sklearn Models API
• Sklearn Regressors API

• The Bias-Variance Tradeoff
• Ensemble Theory
• Regression vs Classification
• Regularisation
• Why Ensembles Win