Exercise 3: What Makes These Two Case Studies Different From Invented Examples — Possible Solution ==================================================================== WHAT THE WARN-BOX SAYS DIRECTLY ------------------------------ Per this chapter's own warn-box, "ds1 spent two entire chapters teaching how to form good, well-motivated questions from data — and then deliberately stopped short of answering any of them, naming that boundary explicitly every time. This course exists specifically to cross that boundary, on the exact same two datasets, so the payoff is concrete rather than abstract." WHAT ds1 ALREADY DID WITH THESE TWO DATASETS ------------------------------ Per this chapter's own earlier list, ds1-9's own used-car listings already went through a full, real seven-step EDA process, ending in a specific, genuinely motivated hypothesis: "does mileage predict price more strongly than year does?" — a question that arose naturally from real patterns actually observed in that dataset's own heatmap and scatter plots, not invented for illustration. Likewise, ds1-10's own employee-attrition table went through the identical seven-step process, producing a specific hypothesis about salary and attrition, plus an explicitly named, still-unresolved confounding-variable question about department vs. salary. WHY THIS IS GENUINELY DIFFERENT FROM AN INVENTED EXAMPLE ------------------------------ A typical invented example is constructed backward from whatever concept the chapter wants to teach — the data exists purely to illustrate the technique, with no real analytical history behind it. These two case studies work the opposite way: the datasets, the patterns within them, and the specific questions being asked all already existed, produced by a real, previously completed analytical process (ds1's own EDA methodology) — before this course, or this particular technique, was ever chosen to answer them. The hypotheses aren't manufactured to fit linear regression or logistic regression neatly; they're genuine open questions this course happens to now have the right tools to close. WHY THIS MAKES THE PAYOFF "CONCRETE RATHER THAN ABSTRACT" ------------------------------ An abstract example teaches a technique in isolation, with no prior stakes attached to the specific answer. These two case studies instead let a reader watch a real analytical thread — opened, investigated, and deliberately left unresolved across two entire earlier chapters — get picked back up and actually resolved, using a technique that has a genuine, specific job to do (predicting price, predicting attrition) rather than a job invented to showcase the technique itself. WHY THIS WORKS AS AN ANSWER ------------------------------ It quotes the chapter's own warn-box directly, traces precisely which real, prior analytical work (ds1's own seven-step EDA process) produced each hypothesis, and explains why reusing genuinely pre-existing, motivated questions — rather than manufacturing new ones to fit the technique — is what makes this course's own two case studies categorically different from an ordinary invented teaching example.