AI-Assisted Analytics Workflows: Intake to Audit Pack

A guided, multi-course journey that builds this skill from the ground up. Work through the chapters in order, or jump to any lesson.

Progress0%

From Ask to Contract: What “Done” Must Mean Before You Touch Data Clarifying Questions That Lock Metric, Grain, and Scope Define the KPI So It Can’t Wiggle: Numerator, Denominator, and Inclusion Rules Turn an AI Draft Into a Human-Owned Analysis Plan Acceptance Tests, Audit Logs, Privacy Redaction, and Escalation Triggers

Prompt Anatomy for Testable Analytics Drafts Plan-First Prompts That Surface Assumptions Before Code Prompts for Verifiable SQL: Constraints, Grain, and Test Queries Prompting for Robust Spreadsheet Formulas and Edge-Case Tests I/O Contracts, Minimal Tests, and a Verification Request Prompt for Pandas

Give AI Just Enough Schema to Draft Safe Extraction Prove the Extraction Matches the Metric Grain Catch Join Multiplication Before It Corrupts Totals Lock Down Filters and Time Windows With Boundary Tests Build a Reconciliation Checklist You Can Reuse Make Transformations Deterministic and Reconcile Before vs After

Select EDA Questions and Segments You Can Actually Prove Build Gold Summary Tables as Your Single Source of Truth Pressure-Test Statistical Interpretations Without Becoming a Stats Course Use an Interpretation Checklist to Catch Denominator and Seasonality Traps Draft Chart Specs and Reconcile Every Visual Back to a Gold Table Write an Evidence-Cited Executive Summary With Privacy Guardrails

A Pipeline That Runs Is Not a Pipeline You Can Trust Make the Environment a First-Class Deliverable (venv + Dependencies)Refactor the Notebook Into a Parameterized Script With an I/O Contract Eliminate Hidden State: Ordering, Seeds, and Time Zones Build Logging Checkpoints That Double as Regression Signals Equivalence Tests: Prove the Refactor Preserved Results When Tests Fail: Tolerances, Silent Refactor Bugs, and the Reproducibility Record

Assemble Front Matter: The Analysis Contract a Reviewer Can Execute Package SQL and Transformations With Executed QA Evidence Make Claims Traceable: Map Dashboard and Narrative to Evidence Compile the Prompt, Assumptions, and Equivalence Evidence Into a Review Trail Run the Privacy Review and Publish an Audit-Pack Index and Self-Critique