关于「实战:雾港牌卡从零到第六周的完整测价」,《冷启动测价与贝叶斯调价》给出的定量结论是什么?

结论:第八章让一副虚构牌卡走完全程:问卷给出可接受区间 RM 65.75 到 RM 154.96,成本底线 RM 80.95 压过问卷下沿;六周 6,600 位访客带来 RM 29,941 的收入,期望后悔值 RM 1936;在三个虚构世界里,按收入学习的最坏后悔值 RM 3491,低于凭问卷固定一个价格的 RM 6732;确认性的显著性检验在这一流量上要 270.1 周,所以收敛靠登记好的决策规则。书中以标注框区分已发表的证据、可运行的计算、原典与作者的假说,每条假说都写明什么结果会推翻它;原典考辨取《管子·国蓄》的轻重敛散与「准平」。

数据口径:价格展示与分群定价的法规只写一般框架,不是法律或税务意见。 书中以标注框区分已发表的证据、可运行的计算、原典与作者的假说,每条假说都写明什么结果会推翻它;原典考辨取《管子·国蓄》的轻重敛散与「准平」。

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: 154.96, 65.75, 80.95

floor RM 80.95; PMC RM 65.75; test band RM 80.95 to RM 154.96
candidates inside the band: [89, 99, 109, 119, 129, 139, 149]
arms (every second candidate, RM 20 apart): [89, 109, 129, 149]
arm stated_buy prior_mean a b prior_rev_per_visitor
890.368 0.0442 1.325 28.675 3.930
1090.321 0.0385 1.156 28.844 4.199
1290.254 0.0305 0.914 29.086 3.932
1490.188 0.0226 0.677 29.323 3.361
prior P(best by revenue)89 0.263, 109 0.281, 129 0.256, 149 0.199
查看完整验算程序
# Chapter 8. p81: From survey bounds and cost floor to four price arms with Beta priors (Mistharbour deck).
# Inputs copied from the p80 output: PMC 65.75, PME 154.96, stated buy shares at the candidate prices.
# Floor = unit cost / (1 - fee - minimum margin), as in Chapter 1 (fee 0.03 assumed, margin 0.55, cost RM 34).
# Prior mean conversion = KAPPA x stated buy share; KAPPA = 0.12 is an assumed stated-to-actual calibration,
# not a measured one. Each arm gets an effective prior sample size N0 = 30 (grade C evidence).
# Prior P(best by revenue) from 20,000 draws, seed 81.
import random

PMC, PME = 65.75, 154.96
COST, FEE, MARGIN = 34, 0.03, 0.55
KAPPA, N0 = 0.12, 30
STATED = {79: 0.368, 89: 0.368, 99: 0.351, 109: 0.321, 119: 0.288, 129: 0.254, 139: 0.225, 149: 0.188}

floor = COST / (1 - FEE - MARGIN)
low = max(floor, PMC)
print("floor RM {:.2f}; PMC RM {:.2f}; test band RM {:.2f} to RM {:.2f}".format(floor, PMC, low, PME))
cands = [p for p in STATED if low <= p <= PME]
print("candidates inside the band: {}".format(cands))
ARMS = cands[::2]
print("arms (every second candidate, RM 20 apart): {}".format(ARMS))
priors = {}
print("arm  stated_buy  prior_mean  a       b        prior_rev_per_visitor")
for p in ARMS:
    m = KAPPA * STATED[p]
    a, b = m * N0, (1 - m) * N0
    priors[p] = (a, b)
    print("{:3d}  {:10.3f}  {:10.4f}  {:6.3f}  {:7.3f}  {:21.3f}".format(p, STATED[p], m, a, b, p * m))
rng = random.Random(81)
wins = dict.fromkeys(ARMS, 0)
for _ in range(20000):
    r = {p: p * rng.betavariate(*priors[p]) for p in ARMS}
    wins[max(r, key=r.get)] += 1
print("prior P(best by revenue): " + ", ".join("{} {:.3f}".format(p, wins[p] / 20000) for p in ARMS))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《零历史数据冷启动测价、范氏价格敏感度边界与贝叶斯自适应调价母典》第 8 章「实战:雾港牌卡从零到第六周的完整测价」(版本 0.1.0). 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-133 · v0.1.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026133,
  author    = {Fan, Yafu},
  title     = {Zero-History Cold-Start Price Testing, Van Westendorp Sensitivity Bounds and Bayesian Adaptive Repricing},
  year      = {2026},
  publisher = {5D Library},
  edition   = {0.1.0},
  note      = {DOI pending}
}

本书其他章节结论