logo
📊
AI Learning

AI 数据分析

用 AI 把数据变洞察

📊AI 数据分析

AI Data Analysis Learning Hub -- Let AI Handle Weekly Reports, EDA & SQL Queries

Why Use AI for Data Analysis?

In data analysis, 80% of time is spent on cleaning and formatting, while the truly valuable insight phase gets compressed. We observed a typical scenario in our teaching: an operations team member spent 3 hours every week on reports -- exporting CSV, writing Pandas scripts for missing values, drawing charts, writing conclusions. After using ChatGPT Code Interpreter, the same workflow took just 25 minutes, and she used the saved time for cross-analysis and A/B test design.

This is the core value of AI data analysis: not replacing the analyst's judgment, but automating mechanical steps like data import, coding, and charting so you can focus on business hypotheses and decision validation.

What This Course Teaches

Data Acquisition & Cleaning
You get a CSV with null values, messy date formats, and Chinese field names -- extremely common in real work. The course walks you through using AI to generate Pandas and SQL cleaning scripts, with emphasis on checking whether AI-generated code silently drops data.
EDA & Visualization
Let AI run descriptive statistics first, then recommend suitable chart types. For order data, AI typically suggests a time-series trend + distribution histogram + regional comparison bar chart combo, but you need to judge which charts are actually useful for business decisions.
SQL Copilot
Paste your table structure and field descriptions to AI and let it write SQL. The course emphasizes: first have AI restate its understanding of the fields before writing SQL -- this step avoids many issues where AI confidently writes a logically incorrect JOIN.
Business Insight Output
Analysis is done -- how do you write conclusions? We use the Finding-Evidence-Recommendation structure, with each finding backed by specific data and chart references.
Automated Weekly Reports
Use Zapier or n8n to set up a scheduled task that automatically runs data every Monday morning, generates a Markdown report, and pushes it to Slack. One student said after setting this up: finally no more rushing reports on Sunday nights.
Security & Review
Anonymize data before uploading -- everyone knows this. But what is easy to overlook: AI-generated SQL may miss filters in WHERE clauses, causing result bias. The course includes a review checklist to help you avoid these pitfalls.

Who Is This For

If you are a data analyst or operations professional who works with Excel and SQL daily, this course can multiply your efficiency on repetitive tasks. Most of our students fall into this group -- they have data thinking but coding is not their strength, and they want to use AI to fill this gap.

Complete beginners can follow along too. ChatGPT Code Interpreter supports natural language interaction -- no coding needed for basic analysis. However, if you are not sure what mean and median are, we recommend spending 1 hour on a statistics primer first for better results.

Tools Used

  • ChatGPT Code Interpreter -- Upload CSV for direct analysis, best for quick EDA
  • Claude Projects -- Handle long documents and complex code generation
  • Gemini -- Screenshot recognition and multimodal analysis
  • Pandas / Polars -- Primary Python data processing libraries
  • DuckDB / BigQuery -- SQL analysis engines
  • Tableau / Looker / PowerBI -- BI dashboards and visualization
  • Zapier / Make / n8n -- Build automation workflows

Course Structure

12 chapters in four stages. The first 3 chapters build foundations (tool selection, data cleaning), the middle 4 chapters cover core analysis skills (EDA, SQL, insight output, automation), and the final 5 chapters cover security review and advanced techniques (multimodal analysis, LLM cross-validation). At a pace of 2-3 chapters per week, expect to finish in about 4-6 weeks.

About JR Academy

JR Academy specializes in AI skills training, covering AI programming, data analysis, Prompt Engineering and more. Course content is continuously iterated based on real teaching feedback, with the goal of helping students apply AI tools in their daily work rather than just knowing about them.

Last updated: March 2026

📊
📈
🔍
SQL
🎯 AI Data Analysis
From CSV/SQL to Business Conclusions

Turn Data into Insights

Upload a CSV and let AI clean data, write code, and generate charts -- you focus on business decisions and validating conclusions. One student used this workflow to compress a 3-hour weekly report into just 25 minutes.

data-analysis.txt数据清洗
$这是订单.csv,字段有 order_id, user_id, amount, created_at。请: 1) 给出数据概览(行数/缺失/异常) 2) 生成 Pandas 清洗代码 3) 输出数据字典

Copy the prompt and paste it into ChatGPT or Claude

📊 Data Analysis Self-Assessment

Evaluate your proficiency across 6 capability areas and get personalized learning recommendations

📥 Data Acquisition & Cleaning

Acquire data from multiple sources, handle missing and outlier values, generate data dictionaries

CSV/Excel Import & Export
Basic SQL Queries
API Data Fetching
Data Cleaning (Missing/Outlier Values)
Data Transformation & Standardization
Data Dictionary Writing
Pandas/Polars Operations
Data Quality Detection

常用 AI 数据分析工具

选择适合你的工具组合,从入门到专业全覆盖

典型业务场景

点击查看详细步骤和可复制的 Prompt 模板

📊

运营周报自动化

从 CSV 到图表结论,30 分钟出报告

每周从数据库导出订单/用户数据,用 AI 生成核心指标趋势图和业务洞察,自动推送到飞书/Slack。

周报自动化流程图:从定时触发到数据导出、AI 分析、报告生成、推送

操作步骤

  1. 导出本周数据为 CSV
  2. 上传到 Code Interpreter,要求数据概览
  3. 生成 3-5 张核心图(趋势/分布/对比)
  4. 按发现-证据-建议结构输出结论
  5. 用 Zapier 定时推送 Markdown 报告
💡 示例 Prompt
这是本周订单 CSV,请:
1) 数据概览
2) 与上周对比
3) 生成 3 张核心图
4) 输出周报结论
推荐工具:Code InterpreterZapierSlack
🔬

A/B 测试解读

提供实验表,输出显著性判断和建议

上传实验组/对照组数据,让 AI 计算转化率差异、统计显著性,给出是否上线的建议。

操作步骤

  1. 准备实验数据(含组别、用户ID、转化标记)
  2. 让 AI 计算各组转化率和置信区间
  3. 判断 p-value 是否显著(通常 <0.05)
  4. 输出结论:上线/继续观察/放弃
  5. 标注假设和风险
💡 示例 Prompt
这是 A/B 测试数据,请:
1) 计算转化率
2) 做显著性检验
3) 给出是否上线的建议
4) 列出风险
推荐工具:ClaudePython
👥

客服/工单分析

文本聚类,提取高频问题

导入客服对话记录,让 AI 聚类主题、提取 Top N 高频问题,生成 FAQ 和改进建议。

操作步骤

  1. 导出客服对话记录为 CSV
  2. 让 AI 做文本聚类,识别主题
  3. 输出 Top 10 高频问题及样例
  4. 生成 FAQ 草稿
  5. 给出产品/流程改进建议
💡 示例 Prompt
这是客服对话记录,请:
1) 聚类主题
2) 输出 Top 10 问题
3) 生成 FAQ
4) 改进建议
推荐工具:ClaudeNotion
🗃️

SQL 助手

提供表结构,生成可执行 SQL

描述数据库表结构和业务问题,让 AI 生成 SQL、解释逻辑、标注假设。

操作步骤

  1. 提供表结构和示例数据
  2. 描述要回答的业务问题
  3. 让 AI 生成 SQL 并解释
  4. 要求标注假设和数据质量风险
  5. 在测试环境验证后再用于生产
💡 示例 Prompt
表结构:orders(...), users(...)。
请统计每个区域过去 30 天的 GMV,输出 SQL 和解释。
推荐工具:ClaudeDuckDBBigQuery
💰

财务报表解读

上传 P&L,生成文字解读

上传利润表/现金流表截图或 Excel,让 AI 提取关键数字、计算变化、生成解读。

操作步骤

  1. 上传财务报表(截图或 Excel)
  2. 让 AI 提取关键指标
  3. 计算同比/环比变化
  4. 生成文字解读和风险提示
  5. 标注数据来源和假设
💡 示例 Prompt
这是本季度 P&L,请:
1) 提取关键指标
2) 与上季度对比
3) 生成解读
4) 风险提示
推荐工具:GPT-4oExcel

Prompt 实验室

选择场景类型,复制 Prompt 开始你的数据分析之旅

数据清洗数据清洗
这是订单.csv,字段有 order_id, user_id, amount, created_at。请:
1) 给出数据概览(行数/缺失/异常)
2) 生成 Pandas 清洗代码
3) 输出数据字典

学习路径 & 最新动态

系统化的学习路径,从入门到精通

常见问题

关于 AI 数据分析学习的常见疑问解答

最简单的原则:上传之前先脱敏。把姓名、手机号、邮箱这些 PII 字段替换成假数据或哈希值,再丢给 AI 分析。如果你处理的是公司内部数据,优先用企业版工具(比如 Azure OpenAI 或 Claude Enterprise),数据不会被用于训练。我们在教学中反复强调一点:让 AI 展示它的计算代码和 SQL,这样你可以复核它到底读了哪些字段,有没有在输出里暴露敏感信息。
不建议。AI 擅长快速出初步结果,但它会犯一些很隐蔽的错误——比如 SQL 里漏了 WHERE 条件导致数据范围偏大,或者用错了聚合方式。我们推荐的做法是:让 AI 输出代码和计算过程,你自己跑一遍确认数字对不对,重要结论再用另一种方法(比如 Excel 手算一个小样本)交叉验证。
可以,但得摆正预期。有个学员之前做行政,完全没写过代码。她用 Code Interpreter 上传 Excel 问「帮我看看哪个月销售额最高」,AI 直接给了图表和答案——这个层面确实不需要写代码。但到了要定制清洗逻辑、调整图表样式的时候,你至少得能看懂 AI 生成的 Python 代码,知道哪里该改。课程前 3 章会带你建立这个基础,不用提前学。
Prompt 大师教的是通用 prompt 技巧,比如角色设定、思维链、few-shot 这些,适用于写作、翻译、代码等所有场景。AI 数据分析这边更垂直——你会学到怎么描述表结构让 AI 写出正确的 SQL,怎么让 AI 生成 EDA 代码而不是泛泛的建议,怎么搭建自动化周报。如果你的日常工作就是跟数据打交道,直接来这个方向。
最直接的变化是分析速度。以前写一个周报要半天,学完之后 30 分钟能搞定——包括数据清洗、画图、写结论。具体来说,你能用 AI 生成清洗代码和 SQL,自动输出带图表的分析报告,还能用 Zapier 搭定时推送。这些技能在运营分析、产品分析、BI 岗位上都用得到。
一个 ChatGPT Plus 账号就能开始。后面章节会用到 Jupyter Notebook 和 DuckDB,课程里有安装指引,跟着走就行。
关键是保留完整的代码和假设记录。让 AI 输出代码而不只是结论,把 SQL、Python 脚本、数据来源、过滤条件都存到 Git 仓库或 Notion 文档里。下次别人要复现你的分析,直接跑代码就行,不用猜你当时是怎么算的。
Excel 处理几千行数据没问题,但到了几十万行就卡了。而且 Excel 的痛点不在计算能力——是重复操作。每周导数据、做透视表、画图、写结论,这套流程用 AI 可以一句 prompt 搞定。另外 AI 能直接生成 Python/SQL 代码,方便版本管理和自动化调度,这是 Excel 做不到的。
能用,但不够精致。Code Interpreter 生成的 matplotlib 图表默认样式比较朴素,直接放到老板的 PPT 里可能不太合适。我们在课程里教了两种解决办法:一是让 AI 输出 Altair/Plotly 交互式图表,样式好很多;二是把 AI 输出的数据导入 Tableau 或 Looker 做最终呈现。初步分析用 AI 出图快速看趋势,正式汇报再用专业 BI 工具打磨。
🎯

Learn More Efficiently

Register to unlock all features

  • ✏️Take study notes
  • 📈Track progress
  • 🏅Get certificate
  • 💬Community Q&A
Sign Up Free →Already have an account? Sign in
👥5,000+ learners enrolled
Register to unlock all features
Notes · Progress Tracking · Certificate
Sign Up Free