Exercise 1: Why Filling Beat Dropping Here, and When Dropping Would Win Instead — Possible Solution ==================================================================== THE TRADE-OFF, PER THIS CHAPTER ------------------------------ Per this chapter, "dropping row 1003 entirely discards a genuine sale that happened, distorting any total revenue figure downstream. Filling it with a placeholder like 'UNKNOWN' keeps the sale in the total but makes any per-store analysis (ds1-5's own groupby) silently wrong for that one row." WHY FILLING WAS JUDGED BETTER FOR THIS SPECIFIC DATASET ------------------------------ Per this chapter, "for this dataset, filling with 'UNKNOWN' is the better trade-off — the revenue total stays accurate, and the ambiguity is at least visible and labeled rather than silently dropped." The decision hinges on which downstream question matters more for this particular dataset: ds1-1's own retail-sales narrative was specifically about total revenue trends (a company noticing weekend sales dipped), which makes preserving the sale's own revenue contribution the higher priority here. Dropping the row would silently understate total revenue by the exact amount of that missing sale — a real, measurable distortion — while filling with "UNKNOWN" costs only the ability to attribute that one sale to a specific store, a narrower, clearly- labeled gap rather than a silent one. WHY A DIFFERENT SITUATION WOULD FAVOR DROPPING INSTEAD ------------------------------ If the actual analysis goal were specifically about per-store performance — for example, comparing store S1's total revenue against store S2's, exactly the kind of analysis ds1-5's own groupby chapter will cover — then a row with an unknown store_id contributes nothing useful to that comparison at all, and worse, an "UNKNOWN" bucket would sit awkwardly alongside the real store totals, potentially confusing the comparison rather than clarifying it. In that situation, dropping the row would be the better trade-off, since the row's own revenue figure is irrelevant to a question that is fundamentally about per-store attribution, not overall totals. WHY THIS IS A GENUINE JUDGMENT CALL, NOT A FORMULA ------------------------------ Per this chapter's own closing warn-box, "a different analyst, with a different downstream question in mind, might reasonably have chosen to drop that row instead." There's no universal rule that makes filling or dropping objectively correct in isolation — the right choice depends entirely on which specific question the cleaned data will be used to answer, which is exactly why the chapter frames this as a real trade- off rather than presenting one option as simply "the fix." WHY THIS WORKS AS AN ANSWER ------------------------------ It explains the specific reasoning the chapter gives for why filling suited THIS dataset's own revenue-focused analysis, and identifies a concrete, genuinely different analysis goal (per-store comparison) where the same missing value would make dropping the better choice instead, showing the decision is context-dependent rather than fixed.