2024.08.08
Panasonic Group People
Helping Blue Yonder Customers Stay Competitive—How Repeated Fine-Tuning of Demand Forecasting Boosts Inventory Accuracy and Lower Overstock: Nico Bartmann
Series:
SCM Solutions
Improving the precision and reliability of forecasting capabilities through evolutions in AI and machine learning
Enhancing supply chain management solutions through big data analytics
Nico Bartmann
Product Development Department
Blue Yonder
Nico Bartmann joined Blue Yonder in 2021 after obtaining both bachelor’s and master’s degrees in computer science at university and graduate school in Germany. When he first started as a data scientist, he was working on inventory forecasting for inventory management systems used in Blue Yonder’s supply chain management (SCM) solutions, as well as machine learning algorithms (calculation logic, methodologies, etc.), at the company’s Karlsruhe office in the southern part of Germany. In this role, he was functioning as both a data engineer and model developer. He is currently a senior data scientist working on fine-tuning*1 the solution’s demand forecasting capabilities.
＊1 Re-training some or all of the pre-learned and tested models, and adjusting their parameters to take on new tasks.
Blue Yonder is the world leader in digital supply chain transformation and has been part of Panasonic since 2021. More than 3,000 global retailers, manufacturers and logistics providers leverage Blue Yonder to optimize their supply chains from planning through order fulfillment.
To learn more about Blue Yonder visit https://blueyonder.com/
At the Heart of SCM Solutions
Contributing to Users Through Forecasting Capabilities
At the heart of Blue Yonder’s latest SaaS*2 solution, Cognitive Demand Planning, is its forecasting capabilities. This technology enables companies using these solutions to use artificial intelligence (AI) and machine learning (ML) to forecast demand for the products they handle. Incorporating hundreds of variables, the technology produces the relevant data while also calculating business impacts and risks. This enables customers to quickly manage orders and purchases based on highly accurate and reliable data forecasting. In addition to increasing sales, optimizing inventory, and reducing waste, the technology is also expected to help build a more resilient*3 supply chain.
＊2 Software as a Service. Cloud server-based software services available to users via internet networks.
＊3 Supply chains that can flexibly respond to fluctuations, etc.
Working on what is “at the heart” of Blue Yonder solutions is very exciting. I believe that retail customers who use our solutions will be able to get and stay ahead of the competition. Recently, Sainsbury’s, a major UK-based retailer, has acknowledged Blue Yonder as a significant part of its success story. It was very inspiring to see the CEO of Sainsbury’s speak during their full-year results call about how the partnership with Blue Yonder and a solution I worked on was a key reason for their success. Sainsbury’s real-time forecasting with Blue Yonder is optimizing sales, reducing waste, and distributing stock when and where it is needed.
Thorough Pursuit of Hidden Data Patterns
But With No Predetermined Methods To Do So
I work with other data scientists*4/planners and users on the customer side to check and verify the precision of the data that is produced in order to fine tune demand forecasting capabilities. I also explain the reasoning behind the predictions and why the model is behaving like it does, tuning the forecasting capabilities by incorporating feedback from customers.
＊4 Scientists who use big data analytics to create new products and services and improve operational processes.
I’m a data enthusiast and so enjoy digging through data to find new patterns. I use my findings to add to and enhance the current ML models, ultimately ensuring the forecasting capabilities will work within the solution and provide accurate predictions.
In recent years, however, market conditions and customer behavior has been constantly changing. For example, even at a store level, patterns are changing constantly. Against this backdrop, there really isn’t a predefined way to find those patterns, so the work is incredibly challenging. But once you do find those patterns there is a sense of excitement that comes with it. We need to constantly keep pushing the model and adding new features to meet our customers’ needs.
In 2022, Blue Yonder formed a partnership with Snowflake, a company based in Silicon Valley in the U.S., and I have huge expectations for the Data Cloud*5 service that it provides. The partnership with Snowflake allows Blue Yonder to integrate Snowflake’s data capabilities into its Blue Yonder Platform. This enables retailers, manufacturers, and logistics service providers to leverage Blue Yonder’s data cloud offerings to access, share and consume live governed data, industry-specific datasets, and data services at scale. Moreover, use of the Data Cloud will allow more reliable creation and fine-tuning of the machine learning models that I work on. This is a game changer*6 for both Blue Yonder and its customers! Making full use of these benefits, we will do everything we can to contribute to the creation of even better solutions.
＊5 A technology that stores huge amounts of data across multiple servers, making the data easier to access when necessary.
＊6 Revolutionary business development that exceeds industry boundaries and smashes conventional processes and rules to bring change to society.
Automating Manual Tasks and Making Demand Forecasting Capabilities Faster and More Agile
While fine-tuning is very important to ensure accurate forecasting, to optimize the system the fine-tuning process must be repeated several times. Simplifying the task is a major challenge, but we have recently made a significant step toward finding a solution.
Blue Yonder hosts something called the Crystal Ball competition, where associates can showcase their ingenious ideas to improve the customers’ and associates’ experiences. In May of this year, two of my colleagues and I submitted an idea for automating the fine-tuning of demand forecasting at this year’s North America and EMEA Crystal Ball competition, and we won!
Moving forward, our goal is to quickly see this solution in action. We have already formulated a concrete plan for how it will be implemented in the Forecasting Service, and so we are in the process of securing funding for the solution then will begin developing it as soon as possible. Through initiatives like this, we will not only make forecasting capabilities faster but also more agile.
The content in this website is accurate at the time of publication but may be subject to change without notice.
Please note therefore that these documents may not always contain the most up-to-date information.
Please note that German, French and Chinese versions are machine translations, so the quality and accuracy may vary.
Share:
