ML-Driven Demand Forecasting & Supply Optimization

Predictive machine learning model to optimize supply chain operations for an FMCG company's instant noodles distribution network.

---

🎯 Problem Statement

An FMCG company distributing instant noodles across multiple warehouses faced:

• Overstocking at certain distribution points, leading to waste
• Stockouts at high-demand locations, resulting in lost sales
• Inefficient supply allocation due to reliance on manual forecasting

The goal was to build a data-driven model that accurately predicts demand at each warehouse and recommends optimal supply quantities.

---

🔍 Approach

1. Exploratory Data Analysis (EDA)

• Analyzed historical sales data across warehouse locations
• Identified seasonal patterns, demand spikes, and outliers
• Visualized distribution of products and warehouse performance

2. Feature Engineering

• Created lag features and rolling averages for time-based patterns
• Encoded categorical warehouse and product variables
• Normalized numerical features for model training

3. Model Building

• Trained and evaluated multiple regression models
• Compared: Linear Regression, Random Forest, Gradient Boosting
• Selected best model based on RMSE and MAE metrics

4. Supply Optimization

• Used model predictions to recommend optimal stock levels per warehouse
• Minimized both overstock cost and stockout penalty

---

📁 Repository Structure

├── EDA.ipynb                          # Exploratory data analysis
├── Model_Building.ipynb               # Model training and evaluation
├── DSE-FT-CHN-MAY24-G8-Interim-Report.docx  # Project report
└── README.md

---

🛠️ Tech Stack

Tool	Purpose
Python	Core language
Pandas & NumPy	Data manipulation
Matplotlib & Seaborn	Visualizations
Scikit-learn	ML model training
Jupyter Notebook	Development environment

---

📊 Key Results

• Built a demand forecasting model with improved accuracy over baseline
• Identified top-performing and underperforming warehouse regions
• Provided actionable supply recommendations to reduce waste and stockouts

---

🚀 How to Run

# Clone the repository
git clone https://github.com/Rajasekaren-S-K/Machine-Learning-Driven-Demand-Forecasting-and-Supply-Optimization-for-Instant-Noodles-Distribution.git

# Install dependencies
pip install pandas numpy matplotlib seaborn scikit-learn jupyter

# Launch Jupyter
jupyter notebook

Open EDA.ipynb first, then Model_Building.ipynb.

---

📚 Learnings

• End-to-end ML pipeline from raw data to actionable insights
• Feature engineering for supply chain time-series data
• Model evaluation and selection for regression problems
• Business framing of ML outputs for non-technical stakeholders

---

Part of my AI/ML learning journey — transitioning into AI engineering.