data-analyst by 404kidwiz/claude-supercode-skills
npx skills add https://github.com/404kidwiz/claude-supercode-skills --skill data-analyst提供商业智能和数据分析专业知识,专注于 SQL、仪表板设计和基于指标的洞察。通过查询优化、KPI 定义和引人注目的可视化,将原始数据转化为可操作的商业智能。
目标: 为高管团队创建一个“销售业绩”仪表板。
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
需求收集
数据准备 (SQL)
WITH sales_data AS (
SELECT
r.region_name,
s.sales_rep_name,
DATE_TRUNC('month', o.order_date) as sales_month,
SUM(o.amount) as revenue,
COUNT(DISTINCT o.order_id) as deal_count
FROM orders o
JOIN sales_reps s ON o.rep_id = s.id
JOIN regions r ON s.region_id = r.id
WHERE o.status = 'closed_won'
AND o.order_date >= DATE_TRUNC('year', CURRENT_DATE)
GROUP BY 1, 2, 3
),
quotas AS (
SELECT
sales_rep_name,
month,
quota_amount
FROM sales_quotas
WHERE year = EXTRACT(YEAR FROM CURRENT_DATE)
)
SELECT
s.*,
q.quota_amount,
(s.revenue / NULLIF(q.quota_amount, 0)) as attainment_pct
FROM sales_data s
LEFT JOIN quotas q ON s.sales_rep_name = q.sales_rep_name
AND s.sales_month = q.month;
可视化设计(概念性)
实施与交互性
质量检查
目标: 识别注册流程中的瓶颈。
步骤:
定义步骤
SQL 分析
SELECT
COUNT(DISTINCT CASE WHEN step = 'landing_view' THEN user_session_id END) as step_1_landing,
COUNT(DISTINCT CASE WHEN step = 'signup_click' THEN user_session_id END) as step_2_click,
COUNT(DISTINCT CASE WHEN step = 'form_submit' THEN user_session_id END) as step_3_submit,
COUNT(DISTINCT CASE WHEN step = 'email_confirm' THEN user_session_id END) as step_4_confirm
FROM web_events
WHERE event_date >= DATEADD('day', -30, CURRENT_DATE);
计算转化率
洞察生成
目标: 在您的 SaaS 产品中嵌入一个“客户使用情况”仪表板,供用户查看。
步骤:
仪表板创建(参数化)
customer_id。WHERE organization_id = {{ customer_id }}。安全性(行级安全)
customer_id 不能被客户端更改。前端集成 (React)
import { EmbedDashboard } from '@superset-ui/embedded-sdk';
useEffect(() => {
EmbedDashboard({
id: "dashboard_uuid",
supersetDomain: "https://superset.mycompany.com",
mountPoint: document.getElementById("dashboard-container"),
fetchGuestToken: () => fetchGuestTokenFromBackend(),
dashboardUiConfig: { hideTitle: true, hideTab: true }
});
}, []);
性能调优
表现形式:
失败原因:
正确方法:
表现形式:
IF/ELSE 和字符串操作逻辑的计算字段。失败原因:
正确方法:
mart_sales),其中预计算了所有计算字段。表现形式:
失败原因:
正确方法:
视觉设计:
数据完整性:
NULL 值(过滤掉或标记为“未知”)。性能:
可用性:
每周安装
92
仓库
GitHub 星标
43
首次出现
2026年1月24日
安全审计
安装于
opencode79
gemini-cli72
codex71
claude-code66
cursor65
github-copilot65
Provides business intelligence and data analysis expertise specializing in SQL, dashboard design, and metric-driven insights. Transforms raw data into actionable business intelligence through query optimization, KPI definition, and compelling visualizations.
Goal: Create a "Sales Performance" dashboard for the executive team.
Steps:
Requirements Gathering
Data Preparation (SQL)
WITH sales_data AS (
SELECT
r.region_name,
s.sales_rep_name,
DATE_TRUNC('month', o.order_date) as sales_month,
SUM(o.amount) as revenue,
COUNT(DISTINCT o.order_id) as deal_count
FROM orders o
JOIN sales_reps s ON o.rep_id = s.id
JOIN regions r ON s.region_id = r.id
WHERE o.status = 'closed_won'
AND o.order_date >= DATE_TRUNC('year', CURRENT_DATE)
GROUP BY 1, 2, 3
),
quotas AS (
SELECT
sales_rep_name,
month,
quota_amount
FROM sales_quotas
WHERE year = EXTRACT(YEAR FROM CURRENT_DATE)
)
SELECT
s.*,
q.quota_amount,
(s.revenue / NULLIF(q.quota_amount, 0)) as attainment_pct
FROM sales_data s
LEFT JOIN quotas q ON s.sales_rep_name = q.sales_rep_name
AND s.sales_month = q.month;
Visualization Design (Conceptual)
Goal: Identify bottlenecks in the signup flow.
Steps:
Define Steps
SQL Analysis
SELECT
COUNT(DISTINCT CASE WHEN step = 'landing_view' THEN user_session_id END) as step_1_landing,
COUNT(DISTINCT CASE WHEN step = 'signup_click' THEN user_session_id END) as step_2_click,
COUNT(DISTINCT CASE WHEN step = 'form_submit' THEN user_session_id END) as step_3_submit,
COUNT(DISTINCT CASE WHEN step = 'email_confirm' THEN user_session_id END) as step_4_confirm
FROM web_events
WHERE event_date >= DATEADD('day', -30, CURRENT_DATE);
Calculate Conversion Rates
Insight Generation
Goal: Embed a "Customer Usage" dashboard inside your SaaS product for users to see.
Steps:
Dashboard Creation (Parameterized)
customer_id.WHERE organization_id = {{ customer_id }}.Security (Row Level Security)
customer_id cannot be changed by the client.Frontend Integration (React)
import { EmbedDashboard } from '@superset-ui/embedded-sdk';
useEffect(() => {
EmbedDashboard({
id: "dashboard_uuid",
supersetDomain: "https://superset.mycompany.com",
mountPoint: document.getElementById("dashboard-container"),
fetchGuestToken: () => fetchGuestTokenFromBackend(),
dashboardUiConfig: { hideTitle: true, hideTab: true }
});
}, []);
What it looks like:
Why it fails:
Correct approach:
What it looks like:
IF/ELSE and string manipulation logic.Why it fails:
Correct approach:
mart_sales) that has all calculated fields pre-computed.What it looks like:
Why it fails:
Correct approach:
Visual Design:
Data Integrity:
NULL values handled explicitly (filtered or labeled "Unknown").Performance:
Usability:
Weekly Installs
92
Repository
GitHub Stars
43
First Seen
Jan 24, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
opencode79
gemini-cli72
codex71
claude-code66
cursor65
github-copilot65
Excel财务建模规范与xlsx文件处理指南:专业格式、零错误公式与数据分析
45,000 周安装
Meta AI Segment Anything Model (SAM) 零样本图像分割完整指南 | 安装使用教程
232 周安装
API模拟框架:开发测试必备的API Mock工具,支持契约验证与并行开发
233 周安装
数据库设计实战指南:规范化、索引优化与零停机迁移策略
228 周安装
零依赖前端幻灯片工具 - 创建动画丰富的HTML演示文稿,无需构建工具
230 周安装
使用 Playwright 编写 tldraw 端到端测试:完整指南与最佳实践
230 周安装
tldraw 单元测试与集成测试编写指南:Vitest 测试框架与 TestEditor 使用详解
234 周安装
Implementation & Interactivity
Quality Check