I developed a Web App Leveraging Machine Learning to Forecast Multi-SKU Product Demand
Managing stock across thousands of SKUs is no joke — especially when dead stock eats into your margins and stockouts kill your customer experience. Predicting product demand isn’t just hard, it’s mission-critical.
🔍 Inspired by a brilliant project by @Abhiram AS (CFO, Seeken Electronics) — thank you for the initial code and the spark of an idea — I built a solution to tackle this challenge head-on.
🎯 The Solution? A demand forecasting tool powered by ARIMA (AutoRegressive Integrated Moving Average) — a statistical model known for its precision in time series forecasting.
🧠 It analyzes 500,000+ rows of historical sales data and generates monthly demand forecasts for each product. This helps avoid overstock, reduce waste, and keep inventory aligned with real-world demand.
🌐 Try it here: https://arima.fuhadabdulla.com/
⚙️ Built and deployed using Dash by Plotly, the app offers a sleek and intuitive interface. (Note: The live version uses truncated data due to server limitations, but it processes full datasets locally.)
📊 Dataset: Based on a Kaggle e-commerce dataset — scalable, flexible, and industry-relevant.
Let’s turn data into action. Let’s solve dead stock before it happens.