AI and ML for Business Insights

Vocabulary terms and definitions covering the landscape of AI, machine learning types, model evaluation, and business applications based on the lecture material.

Artificial Intelligence

The broad field of machines simulating human intelligence.

Machine Learning

A subset of AI where machines learn patterns from data without being explicitly programmed.

Supervised Machine Learning

A subset of Machine Learning where a model is trained on labeled data with known correct answers so it can predict future outcomes.

Unsupervised Machine Learning

A subset of Machine Learning where the model finds patterns and discovers relationships in data with no predefined labels or outcomes.

Decision Tree

A supervised machine learning model that classifies data according to a pre-defined outcome based on the characteristics of that data.

Hallucination

When a Large Language Model (LLM) generates confident-sounding but factually incorrect output.

Association Mining

An unsupervised machine learning technique that finds which events predict the occurrence of other events, most commonly used for market basket analysis.

Lift > 1

A score in Association Mining indicating items are more likely to be purchased together than by chance, denoting a useful rule.

Lift < 1

A score in Association Mining indicating items are less likely to be purchased together, suggesting a negative association.

Lift $= 1$

A score in Association Mining indicating no relationship exists between items.

Large Language Models (LLMs)

Deep learning models that predict the next word in a response based on training data from sources like Wikipedia, books, and articles.

Retrieval-Augmented Generation (RAG)

A technique used to reduce hallucinations in LLMs by grounding responses in verified sources.

Explainability in AI

The ability to understand and explain why an AI model made a specific decision, which is critical for regulatory compliance, trust, and accountability.

Predictive Analytics

A type of analysis used to forecast future outcomes based on past data, suitable for problems like customer churn or fraud detection.