Overview

Traditional dashboards are effective at showing what happened, but they often fall short in explaining why it happened or what actions should be taken.

This project simulates a modern analytics workflow by transforming raw transactional data into decision-ready insights. It combines data engineering, analytics, and AI to automate the reporting process end-to-end.

Key Features

• KPI tracking(GMV, Orders, AOV)

• Anomaly detection using rolling statistics and z-scores

• Cohort-based customer retention analysis

• Customer segmentation(RFM-style)

• Automated insight generation

• AI-generated business reports(LLM integration)

• Human-in-the-loop approval workflow

Tech Stack

• Python

• SQL(DuckDB)

• Pandas

• Streamlit

• Matplotlib

• LLM APIs(OpenAI / Anthropic)

Example Outputs

• KPI trend dashboards

• Anomaly detection signals

• Customer retention cohorts

• Segmentation summaries

• AI-generated weekly reports