Course materials, organized by week. Everything posted here is also available inside Canvas; this page is a direct way to browse it.
-
Week 1 — LLM Context Mastery
Context engineering: giving an LLM the context it needs to work as your analyst.
-
Week 2 — Prompt Writing & Spec Documents
Excel is deterministic; an LLM is not. A careful specification is what makes a non-deterministic tool reliable.
-
Week 3 — Data Sources: AR, Aging, and Cash Application
Build the AR aging report in Excel, then recreate the same headline numbers in SQL on the in-browser SQL Workbench.
-
Week 4 — Excel Fundamentals & FSA Setup
The Excel formula toolkit until it is automatic, then ingest the real Compustat data with Power Query and winsorize it. The workbook you build starts Project 1.
-
Week 5 — Excel Analytics & Visualization
The chart is the deliverable. Age Summit Gear's receivables by days past due, build the six charts that answer the controller, and read each one honestly.
-
Week 6 — Reading Distributions & Telling the Story
No new homework this week. Read the shape of your data, choose mean or median honestly, and turn your Project 1 analysis into a deck a decision-maker can act on.
-
Week 7 — Python + Pandas for Accounting Data
Every Excel move has a Python equivalent. Rebuild the Summit Gear receivables in pandas, where a function is a named formula, a loop is a running total, and a mask is an IF. Once it is code, it re-runs next quarter for free.
-
Week 8 — Predictive Modeling: Regression & Classification
Fit OLS and logistic regression on the Summit Gear customers, taught AI-first: predict the number, then predict the yes/no, and read what each coefficient is really telling you.
-
Week 9 — Forecasting, Part 2: From Ideas to Predictions
Survey the most common forecasting approaches, rank them with a backtest, and hedge the number you report. No new lab or homework; three interactive trainers build the intuition while you finish Project 2.
-
Week 10 — Accounting Automation
Clustering is the last new method, then the lecture bridges into automation and determinism. Build a revenue journal entry from a hundred transaction files, screen the outliers, and let the total recompute. Project 3, the capstone, opens this week.