AI for Data Analytics

Welcome and Orientation

• Who is in the room, where everyone stands on AI, and your first question put to a real dataset inside the first hour

What You Need to Know About AI

• How AI actually works and the three consequences that make up the rest of the course
• Why having no internal test for truth means AI can fail confidently
• Why some data must never be entered into a system you do not control
• Why verification is not optional

What You Need to Know About Data

• Structured and unstructured data and where data comes from
• The four levels of analytics, from what happened to what we should do about it
• The habit of asking who is missing from a dataset before trusting what it says

Where AI Earns Its Place, and Where It Does Not

• Five kinds of work worth handing over to AI
• Three conditions under which you should not reach for AI at all

How to Talk to It

• What a strong prompt contains and which of six prompt patterns fits the moment
• How to iterate without starting over
• Why a long conversation eventually stops seeing the file you uploaded

Data You Do Not Know

• How to orient yourself before working with data you cannot check by eye
• Profile a messy, unfamiliar dataset and distinguish errors from real anomalies
• Clean data transparently, explore it, and audit what the cleaning process removed
• Identify rows or information removed without your approval

Communicating What You Found

• Choose which chart answers which question and make AI justify the chart it selected
• Apply five rules for reading a chart critically
• Use three sentences to turn a finding into a decision
• Complete a mini-analysis that runs the full workflow from beginning to end

Unlocking What Was Out of Reach

• Build a working prediction from a paragraph of plain English and interrogate the result
• Read a forecast for its interval rather than only its line
• Pull meaning from thousands of customer reviews without reading each one individually
• Apply checks that determine whether predictions, forecasts, and text analysis can be believed

The Errors That Survive a First Check

• Identify fabricated figures and other errors that initially appear reasonable
• Recognize trends that reverse when data is split and hidden third factors that affect results
• Identify models that memorized rather than learned
• Use deliberate checks to catch errors that pass a casual review
• Hunt for problems within analyses that initially appear correct

Putting It Together

• Use seven prompts in one conversation to complete the full workflow
• Examine what changes when nobody is reading the intermediate steps
• Understand agents in practical terms, including how most people will encounter them by inheriting rather than building one
• Solve and verify a real question from someone in the room

Capstone and Close

• Apply the workflow to one real task from your own job
• Document the actual prompts used and the verification attached to each
• Identify the first action you will take when you return to your desk

• Students may cancel up to 11 business days before the class/program start date and receive a refund, less the registration fee (10% of the undiscounted course price).
• Cancellations within 11 business days are not permitted, but students may reschedule up to 4 business days before the class start date.

For cancellations and reschedules, please email us at [email protected]