Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/andresshm/fini-marketing-intelligence/llms.txt

Use this file to discover all available pages before exploring further.

Fini Marketing Intelligence is a complete data platform built for candy brand analytics. It generates a realistic synthetic dataset of 20 products, 5,000 customers, and 100,000 sales transactions, loads them into a PostgreSQL data warehouse, and then runs customer segmentation, strategic insight generation, and three-tier revenue forecasting — all from a single pipeline command.

Quickstart

Run the full pipeline in minutes — from data generation to forecast outputs.

Architecture

Understand the star-schema data warehouse and how all components connect.

ETL Pipeline

Explore synthetic data generation, PostgreSQL loading, and validation logic.

RFM Segmentation

Champion, Loyal, Potential Loyalist, and At Risk customer segments explained.

Analytics Queries

Ready-to-run SQL for KPIs, promotions, regions, launches, and seasonality.

Forecasting Models

Compare Prophet baseline, enriched seasonal, and XGBoost forecasting models.

How it works

1

Generate synthetic data

Run the pipeline to create 20 candy products (Gummies, Belts, Seasonal, etc.), 5,000 customers across five regions and four channels, and 100,000 sales with realistic seasonality and discount patterns.
2

Load into PostgreSQL

The ETL module loads all three datasets into a star-schema data warehouse (dim_products, dim_customers, fact_sales) via Docker Compose. Validation confirms row counts before proceeding.
3

Run analytics and segmentation

SQL views (vw_sales_enriched, vw_product_launches) power nine analytics queries. The RFM engine segments customers into Champions, Loyal Customers, Potential Loyalists, and At Risk tiers and writes results back to the customer_rfm table.
4

Forecast revenue

Three models — Prophet baseline, Prophet with seasonal regressors, and XGBoost with lag/calendar features — produce 90-day revenue forecasts with MAE, RMSE, and MAPE metrics exported as JSON.

Key capabilities

Seasonal simulation

Sales data respects product seasonality: Halloween (October), Christmas (December), Summer (June–August), and Valentine windows with configurable probability weights.

Strategic insights

Automatically generated markdown reports covering customer value tiers, promotion effectiveness, top product launches, and underperforming regions.

Power BI integration

A pre-built fini_BI.pbix dashboard connects directly to the PostgreSQL warehouse for interactive exploration of all KPIs.

Reproducible pipeline

Fixed random seeds (seed=42) ensure every run produces identical datasets, making results fully reproducible for development and testing.
Start with the Quickstart to get the full pipeline running, or jump to Architecture Overview if you want to understand the data model first.

Build docs developers (and LLMs) love