Module 1: Foundations of AI · BoK IV.A
Five algorithms to recognise on sight
Governance pros must recognise five algorithms to talk risk: linear regression, logistic regression (binary outcome), decision trees, random forests (an ensemble more accurate than a single tree) and neural networks.
Engineers choose these, but governance pros must understand them to talk risk. Match algorithm → job → example.
| Algorithm | Job | Example |
|---|---|---|
| Linear regression | Numeric predictions from continuous variables | Predicting natural disaster damage costs from housing prices and infrastructure |
| Logistic regression | Probabilistic technique → likelihood of an event, usually a binary outcome | Will this customer stay or churn · will a product appeal to a demographic |
| Decision trees | Supervised learning for classification and regression | Spam vs not-spam email filters |
| Random forests | Ensemble of trees → more accurate, handles complex data better than single decision trees | Recommendation engines |
| Neural networks | Mimics the human brain → highly complex tasks | NLP, facial recognition, ChatGPT, Google Gemini |
Key terms - quick answers
What is “Linear regression”?
Algorithm making numeric predictions from continuous variables.
What is “Logistic regression”?
Probabilistic algorithm giving the likelihood of a usually binary outcome.
What is “Decision trees”?
Supervised learning algorithm for classification and regression.
What is “Random forests”?
Ensemble of decision trees, more accurate and better at complex data than a single tree.