From b9e331cb31c49cf3ad193a6a97451d183b7fbc60 Mon Sep 17 00:00:00 2001 From: cflmy Date: Sun, 18 Oct 2026 02:38:48 +0800 Subject: [PATCH] test(notebook): add Python business-insights example Five-cell Python notebook sharing one kernel: deterministic order generation, regional/product metrics, z-score anomaly detection, ASCII report + JSON, and an isolated second Python kernel that validates the report through an explicit file bridge. --- tests/notebook/python-insights.mq.md | 208 +++++++++++++++++++++++++++ 1 file changed, 208 insertions(+) create mode 100644 tests/notebook/python-insights.mq.md diff --git a/tests/notebook/python-insights.mq.md b/tests/notebook/python-insights.mq.md new file mode 100644 index 0000000..faf3b1d --- /dev/null +++ b/tests/notebook/python-insights.mq.md @@ -0,0 +1,208 @@ +--- +title: Python 经营洞察 Notebook +description: 多 cell 共享 Python 内核,做订单生成、区域/产品指标、异常检测、报告输出与独立复核。 +--- + +# main + +`内核表` = + +| 名称 | 语言 | 命令 | 说明 | +|------|------|------|------| +| 分析 | python | | 共享数据清洗、指标与异常检测 | +| 复核 | python | | 独立内核,只通过显式 JSON 文件复核 | + +`会话表` = + +| 单元 | 内核 | 说明 | +|------|------|------| +| 准备数据 | 分析 | 生成可复现的订单数据 | +| 计算指标 | 分析 | 区域、产品、周度趋势 | +| 检测异常 | 分析 | 按区域金额做 z-score | +| 生成报告 | 分析 | ASCII 图 + JSON 报告 | +| 独立复核 | 复核 | 另一个内核读取报告做校验 | + +`准备数据` = +```python +import math +import random +from collections import defaultdict +from datetime import date, timedelta + +SEED = 20261017 +rng = random.Random(SEED) +START = date(2026, 1, 5) +REGIONS = ["华东", "华南", "华北", "西部"] +PRODUCTS = {"A": "智能传感器", "B": "边缘网关", "C": "数据订阅", "D": "运维服务"} +BASE_PRICE = {"A": 420.0, "B": 980.0, "C": 260.0, "D": 640.0} + +def make_orders(n=180): + rows = [] + for i in range(n): + week = i // 15 + 1 + region = REGIONS[i % len(REGIONS)] + product = "ABCD"[(i * 7 + week) % 4] + seasonal = 1.0 + 0.12 * math.sin(week / 2.2) + qty = max(1, int(rng.gauss(3.0, 1.2))) + amount = round(BASE_PRICE[product] * qty * seasonal * rng.gauss(1.0, 0.08), 2) + rows.append({ + "id": f"O{i + 1:04d}", + "week": week, + "region": region, + "product": product, + "qty": qty, + "amount": amount, + "day": (START + timedelta(days=i * 4)).isoformat(), + }) + return rows + +orders = make_orders() +assert len(orders) == 180 +print(f"生成订单: {len(orders)} 条") +print(f"时间范围: {orders[0]['day']} → {orders[-1]['day']}") +print(f"种子: {SEED}") +print(f"样例: {orders[0]}") +``` + +`计算指标` = +```python +import statistics + +by_region = defaultdict(list) +by_product = defaultdict(float) +by_week = defaultdict(float) +for o in orders: + by_region[o["region"]].append(o["amount"]) + by_product[o["product"]] += o["amount"] + by_week[o["week"]] += o["amount"] + +region_stats = [] +for region, amounts in sorted(by_region.items()): + region_stats.append({ + "region": region, + "orders": len(amounts), + "total": round(sum(amounts), 2), + "mean": round(statistics.mean(amounts), 2), + "median": round(statistics.median(amounts), 2), + }) +region_stats.sort(key=lambda r: r["total"], reverse=True) + +weekly_avg = [round(by_week[w] / 15, 2) for w in sorted(by_week)] +first3 = statistics.mean(weekly_avg[:3]) +last3 = statistics.mean(weekly_avg[-3:]) +trend_pct = round((last3 / first3 - 1) * 100, 2) +top_products = sorted(by_product.items(), key=lambda kv: kv[1], reverse=True) + +total_revenue = round(sum(o["amount"] for o in orders), 2) +metrics = { + "order_count": len(orders), + "total_revenue": total_revenue, + "avg_order_value": round(total_revenue / len(orders), 2), + "trend_pct": trend_pct, + "region_stats": region_stats, + "top_products": [ + {"code": code, "name": PRODUCTS[code], "revenue": round(value, 2)} + for code, value in top_products + ], +} + +print(f"总营收: ¥{total_revenue:,.2f}") +print(f"客单价: ¥{metrics['avg_order_value']:,.2f}") +print(f"周度趋势: {trend_pct:+.2f}%(首 3 周 vs 末 3 周)") +print("\n区域表现(按营收降序):") +for r in region_stats: + print(f" {r['region']} 订单 {r['orders']:>2} 营收 ¥{r['total']:>10,.2f} 中位数 ¥{r['median']:>8,.2f}") +print("\n产品营收:") +for item in metrics["top_products"]: + print(f" {item['code']} {item['name']} ¥{item['revenue']:,.2f}") +``` + +`检测异常` = +```python +by_region_amounts = defaultdict(list) +for o in orders: + by_region_amounts[o["region"]].append(o["amount"]) + +anomalies = [] +for region, amounts in by_region_amounts.items(): + mean = statistics.mean(amounts) + sd = statistics.pstdev(amounts) + if sd == 0: + continue + for o in orders: + if o["region"] != region: + continue + z = (o["amount"] - mean) / sd + if abs(z) >= 1.8: + anomalies.append({ + "id": o["id"], + "region": region, + "product": PRODUCTS[o["product"]], + "amount": o["amount"], + "z": round(z, 2), + }) + +anomalies.sort(key=lambda a: abs(a["z"]), reverse=True) +print(f"异常订单: {len(anomalies)} / {len(orders)}(|z| ≥ 1.8)") +for a in anomalies[:6]: + print(f" {a['id']} {a['region']} {a['product']} ¥{a['amount']:,.2f} z={a['z']:+.2f}") +if len(anomalies) > 6: + print(f" … 其余 {len(anomalies) - 6} 条") +``` + +`生成报告` = +```python +import json +lines = [] +lines.append("=" * 56) +lines.append("Marqdo Python 经营洞察 · 周报") +lines.append("=" * 56) +lines.append(f"订单数: {len(orders)} 总营收: ¥{total_revenue:,.2f} 趋势: {trend_pct:+.2f}%") +lines.append("") +max_total = max(r["total"] for r in region_stats) +for r in region_stats: + width = max(1, round(20 * r["total"] / max_total)) + lines.append(f"{r['region']:<3} {r['total']:>10,.2f} {'█' * width}") +lines.append("") +lines.append(f"异常订单: {len(anomalies)} 最大异常: {anomalies[0]['id'] if anomalies else '无'}") +lines.append("=" * 56) + +report = { + "title": "Marqdo Python 经营洞察", + "order_count": len(orders), + "total_revenue": total_revenue, + "trend_pct": trend_pct, + "regions": region_stats, + "top_products": metrics["top_products"], + "anomaly_count": len(anomalies), +} +report_path = "/tmp/marqdo-python-insights-report.json" +with open(report_path, "w", encoding="utf-8") as f: + json.dump(report, f, ensure_ascii=False, indent=2) + +print("\n".join(lines)) +print(f"\nJSON 报告已写入: {report_path}") +``` + +`独立复核` = +```python +import json +import os + +path = "/tmp/marqdo-python-insights-report.json" +print("独立内核中 orders 可见:", "orders" in globals()) + +if not os.path.exists(path): + print("未找到报告,跳过复核") +else: + with open(path, encoding="utf-8") as f: + report = json.load(f) + region_sum = round(sum(r["total"] for r in report["regions"]), 2) + total_ok = abs(region_sum - report["total_revenue"]) < 0.01 + count_ok = report["order_count"] == 180 + regions_ok = len(report["regions"]) == 4 + print(f"报告订单数: {report['order_count']}") + print(f"分区域合计: ¥{region_sum:,.2f} / 总额 ¥{report['total_revenue']:,.2f}") + print(f"区域数: {len(report['regions'])} 异常数: {report['anomaly_count']}") + print("复核通过:", total_ok and count_ok and regions_ok) +```