General Championship - Data Analytics'25 - IIT Kharagpur

🏆 Award: Second Runner-Up

🏁 Organized by: Technology Students' Gymkhana IIT Kharagpur

📈 Problem Statement by: AnalytixLabs

---

🚀 Overview

This repository presents a comprehensive solution for optimizing marketing investments and strategy for ElectroMart, a consumer electronics retail firm in Ontario, Canada. Developed as part of the General Championship Data Analytics 2025 Hackathon at IIT Kharagpur, our solution integrates advanced statistical modeling, impact analysis, and optimization frameworks to address core business objectives.

---

🗂️ Repository Structure

├── solution/                                   # Folder containing main solution files
│   ├── Exploratory_Data_Analysis.ipynb         # EDA, SHAP-based Impact Analysis
│   ├── Risk_Analysis.ipynb                     # Herfindahl Index, Volatility, Risk Scores
│   ├── Supply_Chain_KPIs.ipynb                 # KPI's
│   ├── Market_Mix_Modelling.R                  # Meta's Robyn MMM Implementation
│   ├── Market_Mix_Optimization_Framework.ipynb # Novel Optimization Model (MISOCP)
├── Problem Statement.pdf                       # Probelm Statement
├── Data.txt                                    # Contains links to raw and processed datasets
├── Report.pdf                                  # Detailed methodology, results & business strategy

---

🛠️ Tech Stack

• Languages: Python, R
• Libraries: Pandas, Numpy, Scikit, Scipy, Pingouin, Seaborn, SHAP, Catboost, XGBoost, GurobiPy, Robyn
• Tools: Jupyter Notebook, Git
• APIs: Meteostat (Weather Integration)

---

📎 Data Sources

• refer data.txt for data download links.

---

📌 Deliverbales Required to be tackled in the Problem Statement

• What drove revenue (Performance Driver Analysis)?
• How much did each marketing lever contribute (Marketing ROI/Impact Analysis)?
• How to allocate next year’s budget optimally?
• Which product categories should be targeted?
• Which marketing channels work best and why?

---

🧠 Detailed Info about our approach

NOTE:- Refer Report.pdf for more comprehensive idea of our approach.

1. 📊 EDA & Impact Analysis

• GMV trend, discount-revenue patterns, order fluctuations
• Weather data integration via Meteostat API
• SHAP + CatBoost for causal Impact Analysis

2. ⚠️ Risk Analysis

• Channel dependency & volatility using Herfindahl-Hirschman Index
• Risk-stratified budget bounding (Low to Very High Risk channels)

3. 📈 Marketing Mix Modelling for budget allocation (Benchmark for our novel approach)

• Used Meta’s Robyn (SOTA MMM tool) in R
• Daily spend interpolation, spline smoothing, ridge regression, saturation effects

4. 🧮 Our Novel Optimization Framework for optimal budget allocation, finding target product category and indicating best marketing channel

• Custom-built Mixed-Integer SOCP model using Gurobi
• Multi-dimensional constraints (budget limits, risk factors, spend shifts) decided based on EDA and Risk Analysis

---

👥 Contributors

This solution was developed with dedication, collaboration, and passion by a team of skilled students from Azad Hall of Residence, Indian Institute of Technology Kharagpur.