AI response
Recommended AI: ChatGPT, Claude
🧭 Goal, restated — "In 3 months, comfortably explore a dataset and pull a few clear findings from it." Honestly: very doable at 1 hour on weekdays if you build from day one rather than only watching lessons.
Stage 1 · Get your hands on data (≈ 2 weeks)
- Goal: open a real spreadsheet/CSV and not feel lost.
- Key concepts: rows vs. columns as records/variables, data types (number/text/date), filter & sort, basic summary (count/average).
- Activity: grab one public dataset you actually care about (e.g., your own expense log) and ask 5 questions of it.
- Checkpoint: you can answer "what's the average and the biggest outlier?" without help.
Stage 2 · Ask questions with code (≈ 3 weeks)
- Goal: do Stage-1 moves with a few lines of code instead of clicking.
- Key concepts: loading a table, selecting columns, grouping, a simple chart.
- Activity: redo your Stage-1 questions in a notebook; one tiny script per question.
- Checkpoint: you can group by a category and plot it from memory of the pattern.
Stage 3 · Tell a small story (≈ 3 weeks)
- Goal: turn numbers into one clear finding someone else gets in 30 seconds.
- Key concepts: choosing the right chart, before/after framing, one-sentence takeaways.
- Activity: make a one-page mini-report from any dataset, with 1 chart + 3 sentences.
- Checkpoint: a friend reads it and can repeat your main point back.
⚠️ Common traps
- Tutorial loop (endless courses, never your own data) → from week 1, every concept is applied to your dataset that day.
- Tool obsession (which software is "best") → pick one and don't switch for 3 months; the thinking transfers anyway.
✅ This week's first 3 tasks
- Pick one dataset you genuinely care about and open it.
- Write down 5 plain-language questions you'd love it to answer.
- Answer just the easiest one by filtering/sorting — no code yet.





