A scout troop has forty leftover camp badges to sell at a one-day rummage sale. A vendor proposes cameras, three months of labeling, and a model that might raise the sell-through guess by a few badges. Camera gear and labeling would cost more than the leftover inventory is worth. Which action should the troop take?
Select an answer to reveal the explanation.
Short Explanation
Think of hiring a film crew to guess how many leftover badges will sell at a one-day rummage sale. Cameras and three months of labeling cost more than the box of badges is worth, so a slightly better guess cannot pay the bill. A bigger network or random sale-day prices do not make that math work.
Full Explanation
Machine learning projects must clear a cost-benefit bar before data collection begins. Declining the camera-and-labeling proposal is correct because three months of capture plus labeling already exceeds the value of the leftover badge inventory, so even a modestly better sell-through guess cannot repay the spend. Approving because "any model is worth more than leftover badges" treats modeling as intrinsically valuable and ignores that the decision payoff is capped by the box of badges. Training a deep neural network on forty examples does not magically recover camera and labeling cost; deep architectures also demand more data and compute, which widens the gap. Reinforcement learning that tries random prices on sale day still burns the upfront data cost and adds uncontrolled pricing risk without fixing the economics. Exam caveat: a stated reason to decline ML is when collection and modeling cost more than the decision is worth, even if the model would be slightly more accurate. Operational check: estimate total data and modeling cost, estimate the maximum dollar value of a better guess against inventory value and one-day sale upside, and stop if cost exceeds benefit before buying gear.