Exercise 1: Why a Decision Tree's Splits Count as Crisp Logic — Possible Solution ==================================================================== WHAT "CRISP" MEANS, PER THIS CHAPTER ------------------------------ Per this chapter, "classical Boolean logic says a statement is fully true or fully false — an element either is or isn't in a set, 1 or 0, nothing between." A crisp rule produces exactly one of two outcomes for any given input, with no partial or in-between result possible. WHAT A DECISION TREE SPLIT ACTUALLY DOES, PER THIS CHAPTER ------------------------------ Per this chapter, "'is salary below $50,000?' draws a hard line: someone earning $49,999 and someone earning $50,001 — practically identical — land in completely different branches, with zero partial credit for how close the second person actually sits to the boundary." A decision tree's own split condition is evaluated as a strict true-or-false test on every single row: the condition is either satisfied or it isn't, and the row is routed entirely down one branch or the other — never partially down both, and never with any recorded degree of "how true" or "how close" the condition was. WHY THIS MEETS THE DEFINITION OF CRISP LOGIC EXACTLY ------------------------------ Applying this chapter's own definition directly: the split condition ("salary below $50,000") functions exactly like a Boolean statement — for any given employee, it's either fully true or fully false, with the routing decision determined entirely by which of those two states holds. There is no third option, no partial routing, and no numeric degree attached to how far above or below the threshold a given salary actually falls. This is precisely the "1 or 0, nothing between" structure this chapter defines as crisp. WHY THE $49,999/$50,001 EXAMPLE MAKES THIS CONCRETE ------------------------------ Two employees who are, in every practical sense, nearly identical (differing by $2 in salary) receive maximally different treatment from the tree — full membership in one branch's own subsequent decisions, zero membership in the other's — purely because they fall on opposite sides of one specific dollar threshold. This stark, disproportionate outcome for two nearly-identical inputs is the concrete signature of crisp logic's own all-or-nothing structure, exactly the kind of boundary behavior this chapter later identifies as a real, documented source of brittleness. WHY THIS COUNTS AS CRISP EVEN THOUGH ml1-7 NEVER USED THE WORD ------------------------------ ml1-7 described the mechanism (a threshold-based yes/no question at each node) without applying this chapter's own vocabulary to it — the underlying behavior was crisp all along, it simply hadn't yet been named as such. This chapter's own contribution isn't introducing a new mechanism, but retroactively applying a precise term (crisp) to a behavior that was already present, unlabeled, in ml1-7's own material. WHY THIS WORKS AS AN ANSWER ------------------------------ It applies this chapter's own definition of crisp logic directly to the mechanics of a decision tree split as ml1-7 described it, and uses the chapter's own $49,999/$50,001 example to show concretely why two nearly identical inputs receive maximally different treatment — the defining signature of crisp, all-or-nothing logic.