Data Analyst · intermediate
AI-Powered Customer Intelligence & RFM Dashboard
A two-page Streamlit dashboard that runs RFM analysis on retail data and uses AI to generate marketing recommendations.
Project overview
Demonstrates your ability to translate raw transactional data into actionable business strategy — the most critical skill for data analysts in the modern AI-driven market.
Real-world use case
E-commerce companies (like Amazon or Shopify sellers) looking to optimize their marketing spend and increase customer lifetime value.
Tools and technologies
- Python
- Streamlit
- Pandas
- Plotly
- OpenAI API
Skills you can demonstrate
Data Cleaning, RFM Modeling, BI, Prompt Engineering
Build roadmap
- Get the dataset
Download the UCI Online Retail II dataset — the classic for RFM, with real-world messiness. Load it into pandas and record row counts before and after every cleaning step. - Clean the data
Handle negative quantities (returns), missing customer IDs, and date parsing. Write down each decision and why; the interview question is always "what did you throw away?" - Calculate R, F, M scores
Per customer: Recency (days since last purchase), Frequency (order count), Monetary (total spend), each scored 1–5 by quintile. Sanity-check the top customers by hand. - Segment customers
Combine scores into named segments — Champions, Loyal, At-Risk, Lost, New — and check that segment sizes and spend make business sense before building anything on top. - Build the Streamlit dashboard
Two pages: an overview with Plotly charts of segment size and revenue, and a drill-down listing the customers in any segment. Deploy it so the link works in an application. - Add the AI recommendation layer
For each segment, have the model draft a tailored campaign idea and offer. Add a regenerate button, and show one example where you edited the output and explain why.
Ways to extend the project
Integrate the OpenAI API to automatically generate personalized email copy or discount offers for each specific customer segment.
How to showcase your work
- GitHub: Clean code, README, clear data pipeline.
- LinkedIn: Video demo of the dashboard with real customer segments.
- Resume: Impact-focused bullets highlighting business insights.