关于「东西方原典考辨:《墨子·小取》「以类取,以类予」与密尔的差异法」,《合成控制与单站改版》给出的定量结论是什么?

结论:字面意义上的孪生站,平均每个面板只有 0.325 个,有至少一个的面板只占 0.260。密尔要求的那种事例,在三十四个站的店里大约四次里有三次根本不存在。前后对比作为同一对象的两个时刻,偏差 +33.64,均方根误差 35.61。单个孪生站的前期误差 9.43,偏差只有 +1.86,均方根误差却有 8.63:它的偏差小,是因为平移去掉了水平差,可它带着自己全部的噪声,单个站的随机起伏原样进入了估计。合成站的前期误差 6.84,偏差 +3.90,均方根误差 5.42,三者最小。合成站用多个站的加权平均摊薄了单个站的噪声,代价是一点插值造成的偏差。

数据口径:在一个三十四站的示意面板上,真实提升设为每周十二单,前后对比报出 37.77,主要误差来自季节;用单纯形约束权重把三十三个未改版的捐赠站加权成合成对照站,估计为 8.83。 两百个面板上,前后对比的均方根误差为 35.61,合成对照为 5.42。

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: 0.260, 0.325, 1.86, 3.90, 33.64, 35.61

literal twins within 10% on 3 traits: mean 0.325 per panel,
share of panels with at least one: 0.260
methodpreRMSPEbiaslift RMSE
before-aftern/a+33.6435.61
single twin9.43+1.868.63
synthetic6.84+3.905.42
查看完整验算程序
# Chapter 7. p71: Mill's method of difference with one twin site.
# Mill asks for two instances with "every circumstance in common save
# one". Panel as in p11 (illustrative, true lift +12), 200 panels.
# (a) How often does any of the 33 donors match the redesigned site
#     within 10 per cent on level, season loading and market loading
#     at once (a literal twin)? (b) Three comparisons: the same site
#     before and after (Mill's two instances in time), the single best
#     pre-fitted donor shifted to the site's level (a twin in space),
#     and the synthetic site (a twin assembled from many donors).
import math
import random
from operator import mul

# 示意面板:每站周订单 = 基准 + 季节载荷 * 季节 + 大盘载荷 * 大盘 + 噪声
T0, POST, NS = 52, 12, 34


def panel(rep=0, n=NS, t0=T0, post=POST, tau=12.0, start=10):
    rng = random.Random(20261113 + 7919 * rep)
    weeks = t0 + post
    mk = [0.0]
    for _ in range(weeks - 1):
        mk.append(mk[-1] + rng.gauss(0.05, 0.25))
    ss = [math.sin(2.0 * math.pi * (start + t) / 52.0)
          for t in range(weeks)]
    ys, cf, pars = [], [], []
    for i in range(n):
        if i == 0:
            mu, ra, rb, sd = 240.0, 0.20, 1.30, 7.0
        else:
            mu = rng.uniform(120.0, 360.0)
            ra = rng.uniform(0.04, 0.22)
            rb = rng.uniform(0.5, 1.7)
            sd = rng.uniform(5.0, 9.0)
        a, b = mu * ra, mu * rb / 20.0
        y = [mu + a * ss[t] + b * mk[t] + rng.gauss(0.0, sd)
             for t in range(weeks)]
        cf.append(y[:])
        if i == 0:
            for t in range(t0, weeks):
                y[t] = y[t] + tau
        ys.append(y)
        pars.append((mu, a, b, sd))
    return ys, cf, pars, ss, mk


def dot(a, b):
    return math.fsum(map(mul, a, b))


def mean(xs):
    return math.fsum(xs) / len(xs)


def rmspe(xs):
    return math.sqrt(math.fsum([x * x for x in xs]) / len(xs))


def gram(ys, t0):
    n = len(ys)
    pre = [y[:t0] for y in ys]
    g = [[0.0] * n for _ in range(n)]
    for i in range(n):
        for k in range(i, n):
            v = dot(pre[i], pre[k]) / t0
            g[i][k] = v
            g[k][i] = v
    return g


def solve(a, b):
    m = len(b)
    r = [row[:] + [b[i]] for i, row in enumerate(a)]
    for c in range(m):
        p = max(range(c, m), key=lambda i: abs(r[i][c]))
        r[c], r[p] = r[p], r[c]
        for i in range(c + 1, m):
            f = r[i][c] / r[c][c]
            if f:
                r[i] = [x - f * y for x, y in zip(r[i], r[c])]
    x = [0.0] * m
    for i in range(m - 1, -1, -1):
        s = math.fsum([r[i][k] * x[k] for k in range(i + 1, m)])
        x[i] = (r[i][m] - s) / r[i][i]
    return x


def eq_qp(g, c, sup):
    m = len(sup)
    a = [[2.0 * g[i][k] for k in sup] + [1.0] for i in sup]
    a.append([1.0] * m + [0.0])
    b = [2.0 * c[i] for i in sup] + [1.0]
    return solve(a, b)[:m]


def simplex_ls(g, c, tol=1e-10):
    # 精确有效集法:min w'Gw - 2c'w,w >= 0,权重和为 1
    j = len(c)
    k0 = min(range(j), key=lambda k: g[k][k] - 2.0 * c[k])
    w = [0.0] * j
    w[k0] = 1.0
    sup = [k0]
    for _ in range(4 * j + 20):
        gr = [2.0 * (math.fsum([g[i][k] * w[k] for k in sup]) - c[i])
              for i in range(j)]
        nu = -math.fsum([gr[k] for k in sup]) / len(sup)
        out = [k for k in range(j) if k not in sup]
        if not out:
            break
        kb = min(out, key=lambda k: gr[k] + nu)
        if gr[kb] + nu >= -tol:
            break
        sup = sorted(sup + [kb])
        while True:
            z = eq_qp(g, c, sup)
            if min(z) > 0.0:
                for k, v in zip(sup, z):
                    w[k] = v
                break
            step = min(w[k] / (w[k] - v) for k, v in zip(sup, z)
                       if v <= 0.0)
            for k, v in zip(sup, z):
                w[k] = w[k] + step * (v - w[k])
            sup = [k for k in sup if w[k] > 1e-12]
            for k in range(j):
                if k not in sup:
                    w[k] = 0.0
    return w


def synth(ys, g, tr, donors):
    gs = [[g[i][k] for k in donors] for i in donors]
    cs = [g[i][tr] for i in donors]
    w = simplex_ls(gs, cs)
    weeks = len(ys[tr])
    gap = [ys[tr][t] - dot(w, [ys[d][t] for d in donors])
           for t in range(weeks)]
    return w, gap


def ratio(gap, t0=T0):
    return rmspe(gap[t0:]) / rmspe(gap[:t0])


def placebo(ys, g, tr, pool, t0=T0):
    # 逐个捐赠站当处理站;处理站本身不进入安慰剂的捐赠池
    out = {}
    for j in pool:
        dn = [k for k in pool if k != j]
        out[j] = synth(ys, g, j, dn)[1]
    return out


if __name__ == "__main__":
    donors = list(range(1, NS))
    twins, est = [], {"before-after": [], "single twin": [],
                      "synthetic": []}
    pre = {"single twin": [], "synthetic": []}
    for r in range(200):
        ys, cf, pars, ss, mk = panel(r)
        p0 = pars[0]
        n = len([d for d in donors
                 if all(abs(pars[d][k] / p0[k] - 1.0) <= 0.10
                        for k in range(3))])
        twins.append(n)
        est["before-after"].append(mean(ys[0][T0:]) -
                                   mean(ys[0][T0 - 12:T0]))
        g = gram(ys, T0)
        w, gap = synth(ys, g, 0, donors)
        est["synthetic"].append(mean(gap[T0:]))
        pre["synthetic"].append(rmspe(gap[:T0]))
        best = None
        for d in donors:
            sh = mean(ys[0][:T0]) - mean(ys[d][:T0])
            gp = [ys[0][t] - ys[d][t] - sh for t in range(T0 + POST)]
            f = rmspe(gp[:T0])
            if best is None or f < best[0]:
                best = (f, gp)
        est["single twin"].append(mean(best[1][T0:]))
        pre["single twin"].append(best[0])
    have = len([n for n in twins if n > 0]) / len(twins)
    print("literal twins within 10% on 3 traits: mean {:.3f} per panel,"
          .format(mean(twins)))
    print("  share of panels with at least one: {:.3f}".format(have))
    print("method          preRMSPE    bias   lift RMSE")
    for k, xs in est.items():
        pr = "{:8.2f}".format(mean(pre[k])) if k in pre else "     n/a"
        print("{:<14s} {}  {:+6.2f}  {:9.2f}".format(
            k, pr, mean(xs) - 12.0, rmspe([x - 12.0 for x in xs])))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《合成控制法与单站改版因果、捐赠站加权重建反事实与安慰剂置换检验母典》第 7 章「东西方原典考辨:《墨子·小取》「以类取,以类予」与密尔的差异法」(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-304 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026304,
  author    = {Fan, Yafu},
  title     = {Synthetic Control and the Causal Effect of a Single-Site Redesign: Rebuilding the Counterfactual from Weighted Donor Sites and Placebo Permutation Tests},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论