Today, we are thrilled to announce the winners of the VN1 Forecasting Datathon.
The VN1 Forecasting Datathon saw an incredible turnout, with almost 1,000 registered participants submitting more than 3,000 entries across two competitive phases. VN1 Datathon was made possible by the support of our esteemed sponsors—Flieber, Syrup Tech, and SupChains.
Reducing forecasting error has become increasingly valuable in today’s market, where supply chain efficiency can directly impact a company’s profitability, customer satisfaction, and competitive advantage. A reduction in forecasting error, as demonstrated by the winners of our VN1 Forecasting Datathon,—it’s a strategic asset with tangible effects on the supply chain and logistics operations.
Better supply chain planning with optimized logistics, purchases, and production planning
- Inventory Optimization: Reducing forecast error means companies can optimize their inventory levels, striking a balance between overstocking and stockouts. This optimization prevents excess holding costs and minimizes the risk of stockouts, ensuring products are available when and where they’re needed.
- Cost Efficiency: Lower forecasting error translates directly to reduced operational costs. Improved accuracy allows businesses to better plan production, minimize wastage, and streamline procurement. This means fewer emergency shipments, less expedited shipping, and a leaner supply chain overall, which is crucial as logistics costs continue to rise globally.
- Improved Customer Satisfaction: In the era of instant gratification, customers expect products to be available and delivered quickly. Accurate forecasting helps maintain the right stock levels and avoid the disappointing “out of stock” message, contributing to better customer experiences and long-term loyalty.
- Better Cash Flow Management: Holding fewer surplus goods frees up capital, allowing businesses to allocate funds to other areas, such as R&D or expansion. A lower inventory burden also means fewer markdowns, improving profit margins and providing companies with greater financial flexibility.
- Agility in a Volatile Market: The global supply chain has faced significant challenges in recent years, from pandemic disruptions to geopolitical issues and fluctuating consumer demands. Improved forecasting enables businesses to respond with agility to these shifts, enhancing resilience in the face of unpredictable market conditions.
Competition Highlights and Winners
The VN1 Datathon challenged participants to explore novel methodologies in forecasting. Collectively, winners will share a total of $20,000 in prizes. After intense competition and remarkable innovations, here are the top winners of the VN1 Forecasting Datathon:
- Jakub Figura & Philip Stubbs - Achieved an error rate of 46.4%
- Justin Furlotte - 46.6%
- Arsa Nikzad - 47.6%
- Antoine Schwartz - 47.7%
- An Hoang - 48.1%
To put these scores in perspective, a statistical benchmark achieved an error of around 80% on this dataset. Our winners displayed an exceptional level of accuracy, as they achieved a forecast value added (FVA) of around 40%.
The innovations showcased by our Datathon winners illustrate how data-driven forecasting can reduce error rates significantly.
Special Recognition: Best Notebook Documentation
A standout contribution came from Olivier Sprangers, who was awarded for his exceptional notebook documentation. Olivier’s work set a high standard, and his notebooks were used extensively by participants. Below are some of his notable contributions:
-
NeuralForecast Starter
Utilizes DeepNPTS to generate predictions with a neural network-based approach, perfect for capturing complex data patterns. -
MLForecast Starter
Uses LightGBM, a popular machine learning model, optimized for fast, accurate forecasting on large datasets. -
StatsForecast Starter
Employs AutoETS, a statistical approach known for its effectiveness in classical time-series forecasting.
These notebooks will be publicly available post-webinar, offering a valuable resource for practitioners and students alike to refine their skills in forecasting.
Join Our Upcoming Webinar
In celebration of these achievements and to dive deeper into the winning strategies, we’re hosting a Webinar on Wednesday, November 13th. You can register here: https://events.teams.microsoft.com/event/e2021207-9598-45b7-8fae-9b023c319e8f@e4fddd24-e7c0-4641-9390-846430093ede
As an added bonus, we’ll be publishing community notebooks. This will provide an invaluable resource for both professionals and students looking to practice and hone their skills in demand forecasting.
A Heartfelt Congratulations
To all participants, thank you for bringing your dedication, creativity, and talent to this competition. The collective work showcased here represents a leap forward in forecasting methodologies, pushing the boundaries of what’s possible in the field.
Congratulations again to our winners, and thank you to everyone who helped make the VN1 Forecasting Datathon a success. We look forward to seeing the continued impact of your work on the future of forecasting!
Meet Our Sponsors
- Flieber: Known for helping brands forecast demand and optimize inventory, Flieber uses real-time data to help businesses meet their inventory needs accurately and efficiently, ultimately reducing stockouts and improving fulfillment rates.
- Syrup Tech: A cutting-edge provider of demand planning solutions, Syrup Tech leverages AI to drive profitability for e-commerce and retail brands, focusing on reducing waste and improving forecasting accuracy in highly dynamic markets.
-
SupChains: Nicolas Vandeput's own firm, SupChains is dedicated to enhancing supply chain efficiency through education, and data-driven forecasting and inventory models, offering insights based on years of hands-on experience in the field.
The challenge was hosted by Nicolas Vandeput, a renowned expert in forecasting and inventory optimization. Nicolas, is acknowledged for his commitment to advancing knowledge in supply chain forecasting. With best-selling titles like Data Science for Supply Chain Forecasting (2018) and Inventory Optimization: Models and Simulations (2020), Nicolas brings a wealth of expertise to the field, making this competition a unique and rewarding experience for participants.
The innovations showcased by our Datathon winners illustrate how data-driven forecasting can reduce error rates significantly. By achieving accuracy as high as 46.4%, these participants are setting new standards for efficiency and resilience in supply chains. In sum, reducing forecasting error and optimizing logistics aren’t just operational goals—they’re strategic priorities that yield financial, operational, and environmental benefits. As supply chains become more sophisticated and customer expectations evolve, the ability to forecast and manage logistics effectively will remain critical to staying competitive in today’s market.
