#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ OPC 城市选址六维加权量化评分引擎 v2.0 ====================================== 依据《中国31城OPC创业选址白皮书》2026 附录C 评分规则复现。 八维权重(v4.0): D1市场0.22 / D2成本0.20 / D3产业0.18 / D4政策0.12 / D5人才0.08 / D6生活0.07 / D7数字基建算力0.08 / D8融资创投0.05 总分公式: total = (D1*0.22 + D2*0.20 + D3*0.18 + D4*0.12 + D5*0.08 + D6*0.07 + D7*0.08 + D8*0.05) * 10 评级: S>=85 / A75-84 / B60-74 / C45-59 / D<45 v4.0 升级（2026-09）: 六维→八维，新增 D7数字基建与算力、D8融资与创投环境（资质型评分，来源：工信部先导区/发改委东数西算/科技部试验区/胡润独角兽）。 用途: ① 用已评分的 D1-D6 重算总分(校验白皮书数据) ② 用原始指标(附录C区间) 计算任意新城 D1-D6 -> 总分 ③ 生成 TopN 排名(按总分排序; 默认全量导出) 2026-08-31 十轮自检重构（缺陷全闭环）: - 删除 grade() 死代码(未定义变量town+if False恒假) 与 _grade_by_total() 中转 - _pick_band 输入防护(非数值/<=0 -> None) + 全区间表去重叠(D1边界双命中偏下已修) - rank() 默认全量(防100截断) + total=None防护 + 同分稳定排序 + 不污染源对象 - normalize() 统一 D 键小写(根除历史"5城大写致六维空白"事故) - validate 支持 31 城/103 城全量双校验 """ import json, sys, os WEIGHTS = {"D1": 0.22, "D2": 0.20, "D3": 0.18, "D4": 0.12, "D5": 0.08, "D6": 0.07, "D7": 0.08, "D8": 0.05} # ===== D7/D8 资质名单（v4.0 八维扩展·来源标注） ===== # 国家人工智能创新应用先导区（工信部·三批） AI_XIANDO = ['北京','上海','深圳','广州','杭州','南京','成都','武汉','长沙','天津','济南','青岛','苏州'] # 东数西算枢纽集群城市（国家发改委2022·103城内的集群城市） DONGSHU = ['呼和浩特','上海','苏州','芜湖','贵阳','成都','重庆'] # 国家新一代人工智能创新发展试验区（科技部） AI_SHIYAN = ['北京','上海','天津','深圳','杭州','合肥','广州','济南','成都','武汉','南京','沈阳','苏州','郑州','西安','重庆','长沙','青岛'] # 独角兽企业所在城市（胡润全球独角兽榜·分档） DUJIAO_T1 = ['北京','上海','深圳','杭州'] DUJIAO_T2 = ['广州','南京','成都','武汉','苏州'] DUJIAO_T3 = ['合肥','西安','长沙','重庆','天津','青岛','厦门','无锡','常州','宁波','嘉兴'] # 创投/引导基金保底 = 省会 + 直辖市 + 计划单列市 + 苏州 SHENGHUI = ['石家庄','太原','呼和浩特','沈阳','长春','哈尔滨','南京','杭州','合肥','福州','南昌','济南','郑州','武汉','长沙','广州','南宁','海口','成都','贵阳','昆明','西安','兰州','西宁','银川','乌鲁木齐','拉萨'] ZHIXIA = ['北京','上海','天津','重庆'] JIHUA = ['大连','青岛','宁波','厦门','深圳'] BAODI = set(SHENGHUI + ZHIXIA + JIHUA + ['苏州']) GANGAOTAI = ['香港','澳门','台湾省'] def d7_score(city): if city in GANGAOTAI: return None s = 2 # 基础分：所有城市均具备互联网/数字基础设施 if city in AI_XIANDO: s += 3 if city in DONGSHU: s += 3 if city in AI_SHIYAN: s += 2 return min(s, 10) def d8_score(city): if city in GANGAOTAI: return None s = 2 # 基础分：所有城市均具备基础金融服务 if city in DUJIAO_T1: s += 3 elif city in DUJIAO_T2: s += 2 elif city in DUJIAO_T3: s += 1 if city in BAODI: s += 5 return min(s, 10) def total_from_d(d): return round((d["D1"]*WEIGHTS["D1"] + d["D2"]*WEIGHTS["D2"] + d["D3"]*WEIGHTS["D3"] + d["D4"]*WEIGHTS["D4"] + d["D5"]*WEIGHTS["D5"] + d["D6"]*WEIGHTS["D6"] + d.get("D7", 0)*WEIGHTS["D7"] + d.get("D8", 0)*WEIGHTS["D8"]) * 10, 1) def grade_from_total(t): if t is None: return None if t >= 85: return "S" if t >= 75: return "A" if t >= 60: return "B" if t >= 45: return "C" return "D" # ---------- 附录C：原始指标 -> 1-10 分映射（区间已去重叠：[lo,hi] 连续无缝隙） ---------- def _pick_band(value, table): """value(数值) 或 None；table 为 [(lo, hi, score)...]，(lo=0 表示 0= 75 and digital_pct is not None and digital_pct >= 40: return 10 if tertiary_pct >= 70 and digital_pct is not None and digital_pct >= 35 and (opc_industries or 0) >= 5: return 9 if tertiary_pct >= 65 and digital_pct is not None and digital_pct >= 30 and (opc_industries or 0) >= 4: return 8 if tertiary_pct >= 60 and (opc_industries or 0) >= 3: return 7 if tertiary_pct >= 55 and (opc_industries or 0) >= 2: return 6 if tertiary_pct >= 50 and (opc_industries or 0) >= 1: return 5 return 4 # D4 政策友好度（四级加总，满分10） def d4_score(reg_days=None, online=None, subsidy=None, incubator=None): s = 0 if reg_days is not None: s += 3 if reg_days <= 1 else (2 if reg_days <= 3 else (1 if reg_days <= 5 else 0)) if online is not None: s += 2 if online == 2 else (1 if online == 1 else 0) if subsidy is not None: s += 3 if subsidy >= 50000 else (2 if subsidy >= 10000 else (1 if subsidy >= 1000 else 0)) if incubator is not None: s += 2 if incubator >= 20 else (1 if incubator >= 5 else 0) return s # D5 人才可用性：取(高校数/在校生)最高分 D5_UNIV = [(0, 3, 5), (4, 7, 5), (8, 14, 6), (15, 24, 7), (25, 39, 8), (40, 59, 9), (60, 1e12, 10)] D5_STUDENTS = [(0, 9, 5), (10, 19, 5), (20, 39, 6), (40, 59, 7), (60, 89, 8), (90, 119, 9), (120, 1e12, 10)] # D6 生活可负担性：房价收入比（缺失处理见方法论页：按报告等级均值推定并标注） D6_HOUSE = [(0, 8, 10), (8.01, 12, 9), (12.01, 16, 8), (16.01, 20, 7), (20.01, 25, 6), (25.01, 30, 5), (30.01, 35, 4), (35.01, 40, 3), (40.01, 1e12, 2)] def d1_score(gdp=None, income=None, retail=None): cands = [_pick_band(gdp, D1_GDP), _pick_band(income, D1_INC), _pick_band(retail, D1_RET)] cands = [c for c in cands if c is not None] return max(cands) if cands else None def d2_score(cost=None, shebao=None): cands = [_pick_band(cost, D2_COST), _pick_band(shebao, D2_SHEBAO)] cands = [c for c in cands if c is not None] return min(cands) if cands else None def d5_score(univ=None, students=None): cands = [_pick_band(univ, D5_UNIV), _pick_band(students, D5_STUDENTS)] cands = [c for c in cands if c is not None] return max(cands) if cands else None def d6_score(house=None): return _pick_band(house, D6_HOUSE) def compute_d(indicators): """从原始指标算出 D1-D6(大写键). indicators 为 dict.""" return { "D1": d1_score(indicators.get("gdp"), indicators.get("income"), indicators.get("retail")), "D2": d2_score(indicators.get("opc_cost"), indicators.get("shebao")), "D3": d3_score(indicators.get("tertiary_pct"), indicators.get("digital_pct"), indicators.get("opc_industries")), "D4": d4_score(indicators.get("reg_days"), indicators.get("online"), indicators.get("subsidy"), indicators.get("incubator")), "D5": d5_score(indicators.get("universities"), indicators.get("students_wan")), "D6": d6_score(indicators.get("house_income_ratio")), } def normalize(d): """统一评分键为小写 d1..d6（引擎输出大写->数据文件小写，导入前必须过此函数； 历史事故：5城 D 键大写导致线上六维空白）。""" out = {} for k in ("d1", "d2", "d3", "d4", "d5", "d6", "d7", "d8"): out[k] = d.get(k) if k in d else d.get(k.upper()) return out def score_ind(indicators): """一键算分: 原始指标 -> {d1..d6, total, grade}（全小写+评级, 与数据文件形状一致）""" dd = compute_d(indicators) total = None if all(v is None for v in dd.values()) else total_from_d(dd) merged = normalize(dd) merged["total"] = total merged["grade"] = grade_from_total(total) return merged def validate_31(path): """重算总分并与白皮书给定值比对.""" data = json.load(open(path, encoding="utf-8")) ok = 0; bad = [] for c in data["cities"]: calc = total_from_d({"D1": c["d1"], "D2": c["d2"], "D3": c["d3"], "D4": c["d4"], "D5": c["d5"], "D6": c["d6"]}) g = grade_from_total(calc) if abs(calc - c["total"]) < 0.051 and g == c["grade"]: ok += 1 else: bad.append((c["city"], c["total"], calc, c["grade"], g)) print(f"校验: {ok}/{len(data['cities'])} 城总分+评级完全复现") if bad: for b in bad: print(" 缺差:", b) return ok == len(data["cities"]) and not bad def validate_all(): """全量校验: 31城复现 + 103城引擎重算与存储值一致.""" c31 = validate_31(os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "data", "opc_cities_31.json")) allp = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "data", "opc_cities_all.json") data = json.load(open(allp, encoding="utf-8")) ok = 0; bad = [] for c in data["cities"]: # 存储值来自引擎/白皮书；仅对含 ind 的城重算比对（31权威城与新城均含 ind） if "ind" not in c: continue dd = compute_d(c["ind"]) if any(v is None for v in dd.values()): continue calc = total_from_d(dd); g = grade_from_total(calc) if abs(calc - c["total"]) < 0.051 and g == c["grade"]: ok += 1 else: bad.append((c["city"], c["total"], calc, c["grade"], g)) print(f"全量校验: {ok}/{len(data['cities'])} 城（含ind且六维齐全）引擎重算一致") if bad: for b in bad: print(" 缺差:", b) return c31 and not bad def rank(limit=None, only="all", out=None): """从数据文件读所有城, 按总分降序生成排名(默认全量导出)。 防护: total=None 排最后; 同分按 d1>d2>... 稳定; 不污染源字典。""" dd = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "data") data = None; src = None for p in [os.path.join(dd, "opc_cities_all.json"), os.path.join(dd, "opc_cities_31.json")]: if os.path.exists(p): data = json.load(open(p, encoding="utf-8")); src = p; break if data is None: raise FileNotFoundError("无数据文件 (opc_cities_all.json / opc_cities_31.json)") cities = [dict(c) for c in data["cities"]] # 拷贝: 不污染源 def sort_key(c): t = c.get("total") return (0 if t is None else 1, -(t if t is not None else 0), -(c.get("d1") or 0), -(c.get("d2") or 0), -(c.get("d3") or 0), -(c.get("d4") or 0), -(c.get("d5") or 0), -(c.get("d6") or 0), c.get("city", "")) cities.sort(key=sort_key) for i, c in enumerate(cities, 1): c["rank"] = i if limit and limit < len(cities): cities = cities[:limit] if out: meta = dict(data.get("meta") or {}) meta.setdefault("count", len(cities)) json.dump({"meta": meta, "cities": cities}, open(out, "w", encoding="utf-8"), ensure_ascii=False, indent=2) print(f"已导出 {len(cities)} 城排名 -> {out}") return cities if __name__ == "__main__": base = os.path.dirname(os.path.abspath(__file__)) datadir = os.path.join(base, "..", "data") data31 = os.path.join(datadir, "opc_cities_31.json") if len(sys.argv) > 1 and sys.argv[1] == "validate": arg = sys.argv[2] if len(sys.argv) > 2 else "all" if arg == "31": validate_31(data31) else: validate_all() elif len(sys.argv) > 1 and sys.argv[1] == "rank": n = int(sys.argv[2]) if len(sys.argv) > 2 else None rank(limit=n, out=os.path.join(datadir, "opc_rank.json")) elif len(sys.argv) > 1 and sys.argv[1] == "score": # 手动打分示例: python3 opc_rank.py score '{"gdp":32039,"income":83436,"opc_cost":2800,"shebao":5000,...}' ind = json.loads(sys.argv[2]) r = score_ind(ind); print("D1-D6:", r); print("总分:", r["total"], "|评级:", r["grade"]) else: validate_all() print("\n用法: validate [31|all] | rank [N] 导出TopN(默认全量) | score '' 给新城打分")