Supply Chain Management — Demand Forecasting with ML

Demand forecasting for a makeup supply chain using a neural network.

🚀 Project Overview

This repository contains a full demand forecasting solution for a fashion & beauty startup’s makeup supply chain. By predicting “Number of products sold”, the model helps optimize inventory, reduce stockouts, and enhance supply-chain efficiency.

Key Steps

1. Data Preparation & EDA
2. Neural-Network Model (128 → 64 → 32 layers)
3. Training & Validation
4. Evaluation (MSE + scatter plot)
5. Model Export & Inference
6. Interactive Streamlit App

📁 Repository Structure

.
├── data/
│   └── supply_chain_data.csv                           # Raw dataset
├── notebooks/
│   └── supply_chain_demand_forecasting.ipynb           # Colab-ready notebook
├── models/
│   ├── demand_forecasting_model.keras                  # Trained Keras model
│   ├── scaler.pkl                                      # StandardScaler used in training
│   └── feature_columns.pkl                             # Feature column order used for prediction
├── streamlit_app/
│   └── app.py                                          # Streamlit dashboard app
├── assets/
│   └── true_vs_predicted.png                           # Sample visualization
└── README.md                                           # Project overview
|__ requirements.txt                                    # All necessary libraries to download.

🛠️ Setup & Usage

1. Install dependencies:

pip install pandas numpy matplotlib scikit-learn tensorflow streamlit joblib

2. Run the notebook:

Open notebooks/demand_forecasting.ipynb in Google Colab or Jupyter:

• Upload your dataset (CSV)
• Clean & preprocess the data
• Train the neural network model
• Evaluate results with MSE & plot
• Save the model, scaler, and feature columns to disk

3. Launch the Streamlit app:

From your project root:

streamlit run streamlit_app/app.py

Upload your CSV to get instant sales predictions and visual feedback!

📊 Streamlit Dashboard Preview

Features:

• Live file upload
• Automatic preprocessing (date + categorical)
• Prediction of "Number of products sold"
• MSE & R² score display
• Scatter plot of true vs predicted
• Option to download prediction CSV

🔄 Next Steps

• 🔧 Perform hyperparameter tuning (e.g., with KerasTuner)
• 📈 Try advanced models like LSTM or XGBoost
• ✨ Enhance Streamlit with interactive filters, export buttons, etc.

---

Built with ❤️ by Anas