SprottSchool of BusinessLearn + frame
Build the method from first principles, connect it to business questions, and define what a valid result would require.
You leave with an evidence target and decision frame.BUSI 4301 · undergraduate AI decision course
Artificial Intelligenceand Business Decision Models
Understand customers. Predict outcomes. Build AI agents. Map markets.
One real marketplace carries you through four cycles: representing customer language, measuring business outcomes, building grounded LLM and agent systems, and analyzing market networks.
One continuous case
4AI decision
cycles
cycles
13complete
classes
4,000 businesses · 72,785 reviews · one connected marketplace
classes
The course journey From data and AI to business decision-making
One decision rhythm across four cycles
First understand it. Then use it. Then test the decision.
Every cycle begins with a business decision, moves into real Yelp evidence, and tests what the result permits a manager to do.
The Cycle Rhythm
Cycles 1 and 2 use three classes; Cycles 3 and 4 concentrate the same analytical rhythm into two.
Execute + interpret
Manipulate the real data and model, inspect what moves, and translate the result into a managerial choice.
You leave with worked evidence and a bounded recommendation.Challenge + decide
Change a consequential assumption, evaluate robustness, and revise or withdraw the recommendation.
You leave with an audited decision and its limits.The four cycles
Four ways business traces become AI evidence.
Select a cycle to see its decision question, data, methods, and what you will produce.
Your course materials
One library-accessible methods stack for four AI cycles.
There is no required textbook purchase. These books are available at no additional cost through Carleton University Library’s institutional O’Reilly access. Brightspace will identify the exact selections for each class.
Required · Cycles 1 and 3Representation, LLMs & Agents
Hands-On Large Language Models
Jay Alammar & Maarten Grootendorst · O’Reilly · 2024
Our main guide to embeddings, text classification, clustering, prompting, retrieval, and controlled language-model systems.
Selected Chapters 1–8, assigned by topicView in O’Reilly ↗Required · Foundations and Cycle 2Business analytics & measurement
Business Analytics with Python
Bowei Chen & Gerhard Kling · Kogan Page · 2025
Supports the business-analytics workflow, Python foundations, data preparation, supervised learning, and model interpretation.
Selected Chapters 1, 4, 5, 7, 8, 10, and 11View in O’Reilly ↗Required · Cycle 4Networks
Network Science
Carlos Andre Reis Pinheiro · Wiley · 2022
Provides the concepts and business applications behind network construction, communities, centrality, similarity, and market structure.
Selected Chapters 1–3 and 5View in O’Reilly ↗Optional advanced extensionGraph Neural Networks
Graph Neural Networks in Action
Namid Stillman & Keita Broadwater · Manning · 2025
An optional bridge from classical network analysis to learned graph representations for students who want to extend the Cycle 4 work.
Selected Chapters 1–2View in O’Reilly ↗How to open the books: sign in through the Carleton University Library, open the O’Reilly database, and search by title. Library authentication is required off campus.
Your semester
Thirteen classes form one decision lifecycle.
Eleven instructional classes build the evidence chain; two presentation classes let teams defend it. Open any class to see what happens, how to arrive, and what you will carry forward.
How your work is assessed
Assessment rewards applied analysis and a defensible business decision.
Assignments build your method practice. The term paper asks you to integrate a business problem, appropriate data, executable AI analysis, interpretation, and evidence boundaries.
10%Attendance
Your analytical development across the four cycles
Each cycle adds a new object to the same evidence discipline.
You will become more independent in specifying what the model consumes, produces, and licenses a manager to decide.
Cycle 1 · RepresentationConstruct analyzable meaning
→
Turn customer language into features and semantic positions, then test whether the representation preserves the distinction the decision needs.
Cycle 2 · MeasurementConnect prediction to action
→
Define targets and timing, compare errors, choose thresholds, and explain what a performance score does—and does not—say.
Cycle 3 · LLMs & AgentsControl an evidence system
→
Specify retrieval, prompts, tools, evaluator, eligibility rules, and stopping behavior so that a fluent answer cannot outrun its evidence.
Cycle 4 · NetworksReason with relationships
Define nodes and edges, interpret communities and centrality, and test whether a market-positioning decision survives another valid graph construction.
How to read any result in this course
Show the chain from raw record to business action.
Begin with the business decision. State what one row, review, score, answer, node, or edge means. Identify what the method constructed. Inspect examples and errors, test a plausible alternative, and only then make the recommendation.
A high score, polished answer, or attractive graph is not the conclusion. Your conclusion is the action the evidence supports under stated conditions—and the condition that would stop you from acting.
1Define the decisionWho acts, on what unit, at what time, and with what cost if the decision is wrong?
2Inspect the dataWhat was observed, selected, omitted, aggregated, or made unavailable?
3Audit the constructionWhat representation, target, prompt, retrieval rule, edge, metric, or threshold created the result?
4Bound the actionWhat survives a plausible alternative, what remains unknown, and when should the manager abstain?
Every strong submission shows this chainBusiness decision → observed data → analytical construction → case evidence → alternative test → bounded action
Before the course begins
Arrive ready to investigate, not already expert.
Your readiness checklist is saved in this browser. The course teaches the AI methods; it expects you to bring a laptop and basic statistical and data literacy.
Questions students usually ask
Know what kind of course you are entering.
By the final class