关于「实证闭环:成交前五题预算问卷、只读题序与预注册检验」,《最少题数的信息论》给出的定量结论是什么?

结论:第八章给出成交前五题预算问卷(示范在第四题停止,区间 3,173 到 8,905)、只读题序、蒸馏小工具「最少题数自适应问卷器」与每组 555 人、共 1,110 人的预注册检验方案。书中以标注框区分已发表的证据、可运行的 Python 计算(19 段程序)、原典与作者的假说。

数据口径:书中的价格、预算、题库与判别力参数均为示意;全部计算由书中程序生成;本书是非个人化经营建议;涉及出生时刻的部分只讨论估计理论与问卷设计,不评价任何体系的预测效度,本书不做任何个人占断。 第四章的模拟中,自适应五题的误差 0.3866 优于固定八题的 0.3969,按精度停止平均只需 4.33 题;信道容量给出任何问法都绕不过的题数下界。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 3173, 8905

stepask priceanswerEAPpost SD80% band
12,948yes4,8390.63712,196 to 11,052
24,783no3,4230.51421,784 to 6,518
33,464yes4,3930.44632,516 to 7,773
44,070yes5,2680.40993,173 to 8,905
stopped after 4 questions; prior SD 0.80 -> 0.4099
查看完整验算程序
# Chapter 8. p81: the five-question budget questionnaire, step by step
# (the core of distill/tool.py). Worked example, parameters illustrative.
# 先验 ln 预算 ~ N(ln 3000, 0.8^2),a = 2.5,题库同 p41(25 个价格,
# 500 到 24,000),每步选当前后验均值处信息最大的价格,
# 后验标准差低于 0.42 或问满 5 题即停。示范回答者:
# 价格不超过 4,200 答「是」,否则答「否」(只为演示,非真实客户)。
import math
from statistics import NormalDist

A, MU, SD = 2.5, math.log(3000), 0.8
BANK = [math.log(500) + i * math.log(48) / 24 for i in range(25)]
GRID = [MU - 4 * SD + 8 * SD * i / 800 for i in range(801)]
STOP_SD, CAP, TRUE = 0.42, 5, 4200


def p_yes(theta, b):
    return 1.0 / (1.0 + math.exp(-A * (theta - b)))


def summary(w):
    m = math.fsum(t * v for t, v in zip(GRID, w))
    s = math.sqrt(math.fsum((t - m) ** 2 * v for t, v in zip(GRID, w)))
    return m, s


def quantile(w, q):
    for i, t in enumerate(GRID):
        if math.fsum(w[:i + 1]) >= q:
            return t
    return GRID[-1]


def money(x):
    return "{:,}".format(int(round(x)))


prior = [NormalDist(MU, SD).pdf(t) for t in GRID]
z = math.fsum(prior)
w = [v / z for v in prior]
m, s = summary(w)
used = []
print("step  ask price  answer  EAP     post SD  80% band")
while len(used) < CAP and s >= STOP_SD:
    b = min((x for x in BANK if x not in used), key=lambda x: abs(x - m))
    used.append(b)
    price = math.exp(b)
    yes = price <= TRUE
    w = [v * (p_yes(t, b) if yes else 1 - p_yes(t, b))
         for t, v in zip(GRID, w)]
    z = math.fsum(w)
    w = [v / z for v in w]
    m, s = summary(w)
    print("%-5d %-10s %-7s %-7s %.4f   %s to %s"
          % (len(used), money(price), "yes" if yes else "no",
             money(math.exp(m)), s, money(math.exp(quantile(w, 0.1))),
             money(math.exp(quantile(w, 0.9)))))
print("stopped after %d questions; prior SD %.2f -> %.4f" % (len(used),
                                                           SD, s))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《费希尔信息矩阵、克拉美罗方差下界与预算及出生时刻探询最少题数母典》第 8 章「实证闭环:成交前五题预算问卷、只读题序与预注册检验」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-235 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026235,
  author    = {Fan, Yafu},
  title     = {Fisher Information Matrices and the Cramér-Rao Bound: the Fewest Questions Needed to Estimate a Budget or a Birth Time},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论