📦 Supply Chain Data Modelling Streamlit App

This Streamlit app is designed to provide advanced analytics and data modelling for supply chain management in the pharmaceutical industry.

With this app, users can explore forecasting, inventory optimisation, customer & product segmentation, and statistical analysis to enhance decision-making in supply chain operations.

---

🏗️ Project Structure

SupplyChainDataModellingStreamlit
├── .streamlit/
│   └── config.toml
├── assets/
│   └── data/
│       ├── DataGenerator.ipynb
│       └── pharmaceutical_supply_chain.csv
├── streamlit_app/
│   ├── pages/
│   │   ├── 01_1️⃣_Forecast_Demand.py
│   │   ├── 02_2️⃣_Inventory_Optimisation.py
│   │   ├── 03_3️⃣_Customer_Product_Segmentation.py
│   │   └── 04_4️⃣_Hypothesis_Testing.py
│   └── 00_ℹ️_Info.py
├── LICENSE
├── README.md
└── requirements.txt

---

🛠️ Features of the App

The application consists of four key modules, each designed to address specific supply chain challenges:

1️⃣ Forecast Future Demand (LSTM Model)

• What does it do?
  • Uses an LSTM (Long Short-Term Memory) neural network to forecast future demand based on historical sales data.
  • Allows users to select a specific SKU and generate a one-year demand forecast.
• How does it work?
  • The dataset is preprocessed to remove non-stationarity and smooth variations in demand.
  • A train-test split (80%-20%) ensures proper model validation.
  • The LSTM model learns from historical sales data and predicts future trends.
• Output:
  • A dynamic time-series forecast plot showing historical sales, predicted test values, and future forecasts.
  • Mean Absolute Percentage Error (MAPE) to measure model accuracy.
  • A downloadable table of forecasted values for further analysis.

2️⃣ Inventory Optimisation & Simulation

• What does it do?
  • Helps users optimise inventory management using:
    • Economic Order Quantity (EOQ) – Determines the ideal order quantity to minimise costs.
    • Safety Stock Calculation – Ensures sufficient buffer stock to prevent stockouts.
    • Monte Carlo Simulation – Simulates demand variations using random sampling.
• How does it work?
  • Users can select an SKU and view its optimal inventory parameters.
  • The Monte Carlo simulation generates 1000+ demand variations to predict uncertainty in sales.
• Output:
  • EOQ & Safety Stock values displayed in a table.
  • Simulated demand distribution to help estimate variability in sales.
  • Interactive visualisation to analyse demand trends and supply chain risks.

3️⃣ Customer & Product Segmentation

• What does it do?
  • Provides segmentation insights based on sales performance and customer behaviour.
  • Users can explore:
    • Sales by Supplier – Identify top-performing suppliers.
    • Sales by Product Family & Category – Discover best-selling products.
    • Top & Bottom Customers – Rank customers by revenue contribution.
    • Stock Turnover Analysis – Understand inventory movement.
• How does it work?
  • Aggregates sales, revenue, and stock turnover across suppliers, categories, and products.
  • Displays interactive bar charts to visualise performance metrics.
• Output:
  • Top & Bottom 20 products by sales & stock turnover ratio.
  • Supplier-wise sales trends to assess reliability & contribution.
  • Customer segmentation based on revenue contribution.

4️⃣ Statistical Hypothesis Testing

• What does it do?
  • Conducts five key statistical tests to validate supply chain insights:
    1. Stockouts reduce sales – Pearson correlation between stock level & sales.
    2. Longer lead times lower sales – Linear regression between lead time & sales.
    3. Supplier revenue varies significantly – Kruskal-Wallis test on supplier revenues.
    4. Frequent buyers generate more revenue – Spearman correlation between purchase frequency & revenue.
    5. Certain categories generate higher revenue – ANOVA test across product categories.
• How does it work?
  • Users can select a hypothesis from a dropdown menu.
  • The app performs statistical calculations and plots the test distribution, rejection region, and observed test statistic.
• Output:
  • Test results with p-value comparison to α = 0.05.
  • Decision to reject or fail to reject the null hypothesis.
  • Visual representation of the test statistic and rejection zone.

---

🎯 Why Use This App?

✔️ Realistic Data Analysis: Built using a synthetic pharmaceutical supply chain dataset.
✔️ End-to-End Workflow: Covers forecasting, inventory planning, customer segmentation, and validation tests.
✔️ Business-Driven Insights: Helps identify demand trends, customer behaviour, and supplier performance.
✔️ User-Friendly Interface: Designed for interactive data exploration and decision-making.

🚀 Get started by selecting a module from the sidebar!