Master Data Analysis with ChatGPT (in just 12 minutes)
About this video
Jeff Su teaches the DIG framework — Description, Introspection, Goal Setting — learned from a Coursera AI for data analysis course as a memorable version of exploratory data analysis (EDA), to turn ChatGPT into a personal data analyst with no technical skills. Using a free Apple TV Plus dataset (titles like Forrest Gump, Avatar: The Last Airbender, The Godfather, Sherlock), Description prompts list all columns with a sample, take five more random samples to spot outliers and inconsistencies, then run a data quality check for missing values, odd formats and outliers — e.g. the title column missing 589 values (3.1%) and available countries missing 99.7%, so no geographic analysis is possible. Introspection has ChatGPT propose 10 answerable questions, show which columns each needs and whether data suffices, and list questions that cannot be answered due to missing viewing, budget or revenue fields. He also shows joining a second dataset via the IMDb ID to add viewership and cost. Goal Setting gives ChatGPT a mission such as what Apple TV should invest in next, producing a roadmap (clean data, genre scorecard, rank opportunities, trend velocity, stress-test outliers) and example insights, plus a final prompt asking what questions readers will challenge.
Key takeaways
- Use DIG in order: Description to understand the file, Introspection to brainstorm and validate questions, Goal Setting to focus analysis on a decision.
- Start Description with column list + one sample, then five more random samples, then an explicit data quality check for missing values, formats and outliers before analysing.
- Let data gaps set expectations: in the demo, 99.7% missing available-countries values ruled out geographic analysis, and missing viewership/budget blocked ROI questions.
- In Introspection, force ChatGPT to name the exact columns needed for each question and whether current data is sufficient, and to list unanswerable questions.
- Before presenting, ask ChatGPT what key questions a reader/manager will ask and how to proactively address them.
Episode details: published July 15, 2025 · 11:54 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are Jeff Su’s own claims, shared for education — they are not typical or guaranteed.
← Previous: The CORRECT way to use ChatGPT (in 2026) · Next: My Brutally Honest Advice About Joining Google (after 9 years) →
All Jeff Su videos · Jeff Su profile · Jeff Su (official site)





