# AI Data-Analysis Prompt Cheat-Sheet — "Talk to Your Data" Workshop
Copy-paste these into **ChatGPT** (chat.openai.com) or **Google Gemini** (gemini.google.com) after uploading a CSV. Free accounts work. Say the file name if the tool asks.

---
## ⭐ THE LIVE DEMO (Module 4) — use `placement.csv`
> Here is my data [upload placement.csv]. Analyse it, show me the top 3 trends as charts, and tell me which factors most affect whether a student gets placed. Then predict whether a student with CGPA 7.5, 1 internship and 3 projects gets placed. Explain it in simple words.

## ⭐ HANDS-ON #1 (Module 5) — students run this on `placement.csv`
> You are helping a beginner. Analyse this student-placement data. In simple words: (1) how many students got placed, (2) the 3 things that matter most for placement, (3) one clear chart. No jargon.

## ⭐ HANDS-ON #2 (Module 7) — prediction + forecast
> Predict whether a student with CGPA [ __ ], [ __ ] internships and [ __ ] projects will get placed, and tell me how confident you are and why.

> Using sales.csv, show the monthly revenue trend as a line chart and forecast the next 3 months. Which product and region are growing fastest?

---
## GENERAL DATA-ANALYSIS PROMPTS (Reference Guide, Part C)
1. **Summarise:** "Give me a one-paragraph summary of this dataset — what each column is and anything surprising."
2. **Top drivers:** "Which columns most strongly influence [target column]? Rank them and show a bar chart."
3. **Clean it:** "Check this data for missing values, duplicates and impossible values (e.g. negative CGPA). Show me what you'd fix."
4. **Chart it:** "Draw the 3 most useful charts for this data and tell me what each one reveals."
5. **Compare groups:** "Compare placed vs not-placed students across every column. Where's the biggest difference?"
6. **Outliers:** "Find the unusual rows in this data and explain why each is unusual."
7. **Correlation:** "Show a correlation heatmap of the numeric columns and explain the strongest relationship in plain English."
8. **Predict (classification):** "Build a simple model to predict [Yes/No column] and tell me its accuracy and the top features."
9. **Predict (number):** "Using housing.csv, predict the price of a 3-bedroom, 1200 sqft house in Kothrud, 5 years old."
10. **Forecast:** "Spot the trend in this time-series and project it forward 3 periods. State your assumptions."
11. **Explain a concept:** "Explain [p-value / overfitting / RAG] to a complete beginner in 4 sentences with one example."
12. **Segment:** "Group these customers into 3–4 meaningful clusters and describe each group."
13. **A/B check:** "Is the difference between group A and group B statistically real, or could it be chance?"
14. **Feature idea:** "Suggest 3 new columns I could calculate from these to improve a prediction."
15. **Sanity check (ALWAYS):** "What could be wrong or misleading in this analysis? What should I double-check before trusting it?"

> **Golden rule for students:** AI gives you a fast first draft — *always verify* the key numbers before you trust them.