Reusable thinking tools
Choose a framework for the task: developing your capabilities, framing a problem, investigating evidence, or checking a result. The two SPARK versions serve separate purposes.
Framework directory
Skills, Portfolio, AI usage, Relationships, and Knowledge in Action. Build useful capabilities, show evidence of your work, use AI thoughtfully, learn with others, and apply knowledge to create value.
Try it this week: choose one skill, create one portfolio artifact, document how AI helped, learn from one person, and explain the result.
Self, Position, Augment, Relationships, and Knowledge to Proof. Inventory what you can demonstrate, identify where it creates value, explore AI support, build relationships, and choose evidence that proves growth.
In Week 1, students create an evidence inventory and a capability target to revisit in Week 10. This is a separate version of SPARK.
Translate the problem, Retrieve evidence, Analyze, Challenge, then Explain and Execute. E includes both communication and action. Keep a visible evidence record throughout.
Organize, Review, Build, Inspect, Tell. Preserve source data, review fields and quality, build the analysis, inspect the results, and tell the decision-relevant finding.
Define the audience, decision, evidence, constraints, and action before starting analysis.
Use AI for acceleration while keeping human responsibility for the source, logic, result, and consequence.
Understand the distinct roles of Excel, SQL, Tableau, Power BI, Python, and AI assistants instead of expecting one tool to do everything.
Use hypotheses, a single source of truth, reasonableness checks, documentation, timeboxes, and the “So What?” test.
Before the software
A short blueprint prevents a technically correct project from answering the wrong question.
Write one or two sentences for each prompt. If an answer is missing, that is useful information: it tells you what to clarify before spending time on analysis.
AI use
Large language models are useful for explanation, drafting, comparison, and critique. They can also produce confident errors.
Role: Who should the AI help?
Context: What audience, goal, data, and constraints matter?
Task: What one concrete output is needed?
Format: How should the answer be organized?
Check: Ask for assumptions, uncertainties, and verification steps.
Compare formulas and queries with documentation. Test a small sample manually. Confirm that sources exist and actually support the claim. Look for missing context, biased assumptions, and conclusions that go beyond the available data.
If the output could affect a grade, business decision, or person's opportunity, a human must inspect the evidence.
Software stack
| Need | Best starting tool | What it does well | Common beginner mistake |
|---|---|---|---|
| Inspect and clean a small file | Excel | Visible cells, approachable formulas, quick summaries, and PivotTables. | Changing cells manually without documenting the steps. |
| Retrieve records from related tables | SQL | Precise, repeatable questions and reliable filtering at scale. | Assuming a query is correct because it ran without an error. |
| Explain a pattern visually | Tableau | Interactive charts, filters, and dashboards for a defined audience. | Adding more charts instead of clarifying the message. |
| Brainstorm, explain, draft, or critique | AI assistant | Fast iteration, plain-language explanation, and alternative perspectives. | Treating plausible text as verified evidence. |
| Model and share business measures | Power BI | Power Query, a data model, measures, and interactive reporting. Optional extension beyond the core course. | Mixing row grain or using an unapproved public sharing option. |
| Repeat cleaning and checks across files | Python | Readable scripts for data preparation, calculations, tests, and repeatable analysis. Optional extension. | Running unfamiliar code without understanding input, output, and permissions. |
Try the public TrailPeak lab, choose an analyst resource, or browse the complete project directory.
