Retail Pricing and Customer Intelligence Suite

This repository features a dual-engine analytics platform that transforms raw e-commerce transaction data into actionable business strategies. The project implements Log-Log Regression for price elasticity and behavioral quintile binning for customer segmentation.

---

Project Overview

The suite provides high-fidelity decision support for retail stakeholders:

• Pricing Intelligence: Identifies price-sensitive vs. inelastic products to optimize profit margins.
• Customer Segmentation: Clusters the user base into actionable cohorts to drive retention.

---

Core Analytics Engines

1. Price Elasticity and Market Audit

The system evaluates product-level demand sensitivity by filtering for statistical significance ($p$-value $< 0.05$) to ensure decision reliability.

• Log-Log Regression: Used to calculate constant elasticity coefficients for 54 statistically significant products.
• Automated Decision Logic: Translates coefficients into strategies such as "Discount for Volume" or "Optimize Margin".

---

2. Customer RFM Segmentation

Utilizing Recency, Frequency, and Monetary metrics, the platform segments customers to resolve the "One-Timer" data bottleneck common in e-commerce datasets.

• Cohort Composition: A Treemap visualization identifies the distribution of the population across five primary segments.
• Prescriptive Actions: Maps specific interventions, such as personalized win-back campaigns for "At Risk" users and premium support for "Big Spenders".

---

Technical Implementation

Data Pipeline Architecture

• Ingestion: Automated download and path-mapping of the Olist Brazilian E-Commerce dataset.
• Processing: Consolidation of orders, items, and payments datasets into a unified transaction log.
• Modeling: Implementation of R and M quintiles to prioritize high-value users in low-frequency environments.

Project Structure

Component	Description
data/raw_orders_data.csv	Consolidated transaction log
data/final_market_audit.csv	Elasticity model outputs
data/rfm_segments.csv	Behavioral segmentation results
src/dashboard.py	Streamlit UI logic
src/models.py	Regression and RFM logic

---

Strategic Insights

• Revenue Optimization: The dashboard identifies products where price increases will not significantly impact volume, protecting margins.
• Customer Retention: Isolated "At Risk" cohorts with an average inactivity period of ~480 days for targeted re-activation.
• Growth Engineering: Identified "New Customers" with high recency (avg. 50.2 days) as the primary targets for loyalty conversion.