关于「投标价格:一个棚日的价值随预订期变化」,《空间租赁收益管理与档期装箱》给出的定量结论是什么?
结论:第三章用随预订期变化的投标价格处理交错到达:投标价格每月比先到先得多收 6009.0 元,而按全月总需求定死的 EMSR-b 只多收 1236.2 元。
数据口径:书中的棚日、价格、需求与到达概率均为示意;全部计算由书中程序生成;本书是非个人化经营建议,所有价格明码标价,稀缺提示只来自真实档期,小工具给出的价格表须经营者人工复核后发布。 第一章在一个 20 个棚日的示意月份里,用精确期望算出保护 10 个棚日可使月收入从 47323.9 元升到 57432.1 元,多 21.36 %,但模拟显示保护在 0.1900 的月份里少收钱。
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: 1236.2, 6009.0
| EMSR-b protection levels on total demand: [2, 7, 14] | ||||
| policy | mean revenue | gain vs FCFS | 95% CI of the gain | |
| DP | 62768.3 | 6009.0 | 5777.3 to | 6240.7 |
| DP-30% | 61697.5 | 4938.1 | 4768.2 to | 5108.1 |
| DP+30% | 61725.8 | 4966.4 | 4672.2 to | 5260.6 |
| EMSR-b | 57995.6 | 1236.2 | 953.8 to | 1518.7 |
| FCFS | 56759.3 | 0.0 | 0.0 to | 0.0 |
查看完整验算程序
# Chapter 3. p32: bid prices, static EMSR-b and FCFS on mixed arrivals.
# Requests arrive one period at a time in the order the model of p31
# draws them (cheap mostly early, dear mostly late, but mixed). Policies:
# DP accept class j iff f_j >= bid price B(t, x) of p31
# DP-30% / DP+30% the same rule built on a forecast in which classes
# 1 and 2 are 30 % lower / higher than the truth
# EMSR-b static nested limits from total class demand (sd = sqrt(mean))
# FCFS accept while any day is left
# 4000 seeded months; paired differences against FCFS with 95 % CI.
import math
import random
import statistics
from statistics import NormalDist
C, T = 20, 100
FARE = [5600, 4200, 3000, 1800]
EARLY = [0.00, 0.02, 0.08, 0.16]
LATE = [0.06, 0.08, 0.04, 0.04]
def probs(t, scale=1.0):
p = EARLY if t > 50 else LATE
return [p[j] * (scale if j < 2 else 1.0) for j in range(4)]
def bids(scale):
v = [[0.0] * (C + 1)]
for t in range(1, T + 1):
prev, row = v[-1], [0.0]
p = probs(t, scale)
stay = 1.0 - math.fsum(p)
for x in range(1, C + 1):
terms = [stay * prev[x]]
terms += [p[j] * max(FARE[j] + prev[x - 1], prev[x])
for j in range(4)]
row.append(math.fsum(terms))
v.append(row)
return [[0.0] + [v[t - 1][x] - v[t - 1][x - 1] for x in range(1, C + 1)]
for t in range(1, T + 1)]
def emsr_b():
mu = [math.fsum([EARLY[j] * 50, LATE[j] * 50]) for j in range(4)]
y = [0]
for j in range(1, 4):
s = math.fsum(mu[:j])
fb = math.fsum(FARE[k] * mu[k] for k in range(j)) / s
y.append(round(NormalDist(s, math.sqrt(s)).inv_cdf(
1 - FARE[j] / fb)))
return y
def run(stream, rule):
x, rev = C, 0
for t, j in stream:
if x > 0 and rule(t, x, j):
x -= 1
rev += FARE[j] # step: integer fare
return rev
B = {s: bids(s) for s in (1.0, 0.7, 1.3)}
Y = emsr_b()
rules = {
"DP": lambda t, x, j: FARE[j] >= B[1.0][t - 1][x],
"DP-30%": lambda t, x, j: FARE[j] >= B[0.7][t - 1][x],
"DP+30%": lambda t, x, j: FARE[j] >= B[1.3][t - 1][x],
"EMSR-b": lambda t, x, j: x > Y[j],
"FCFS": lambda t, x, j: True,
}
CUM = {flag: [math.fsum(probs(t)[:j + 1]) for j in range(4)]
for flag, t in ((True, 100), (False, 1))}
rng = random.Random(20261113)
res = {k: [] for k in rules}
for _ in range(4000):
stream = []
for t in range(T, 0, -1):
u = rng.random()
for j, cp in enumerate(CUM[t > 50]):
if u < cp:
stream.append((t, j))
break
for k, rule in rules.items():
res[k].append(run(stream, rule))
print("EMSR-b protection levels on total demand: %s" % Y[1:])
print("policy mean revenue gain vs FCFS 95% CI of the gain")
for k in rules:
d = [a - b for a, b in zip(res[k], res["FCFS"])]
m = statistics.fmean(d)
h = 1.96 * statistics.stdev(d) / math.sqrt(len(d)) if any(d) else 0.0
print("%-8s %10.1f %10.1f %8.1f to %8.1f"
% (k, statistics.fmean(res[k]), m, m - h, m + h))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《空间租赁动态收益管理与档期装箱、舱位控制嵌套保护水平与碎片空置消除母典》第 3 章「投标价格:一个棚日的价值随预订期变化」(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-282 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026282,
author = {Fan, Yafu},
title = {Dynamic Revenue Management and Slot Packing for Space Rentals: Nested Protection Levels and the Elimination of Fragmented Idle Time},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}