Putting Big Data to Work in Retail - Increasing Margins, Avoiding Food Waste

DDBD Ballroom CD
Average rating: *****
(5.00, 1 rating)

Putting Big Data to work in real-world enterprises is a tricky endeavor – beyond data science and technology. This case study delves into details on how to succeed.
Picking the Right Scenario
Companies around the globe are looking for easy wins out of Big Data applications. The majority knows that they have a lot of data. But most do not know which Questions to ask. Scenarios that sit at the heart of the business promise the biggest returns, but unfortunately they are also the most risky (or perceived as risky). Teaming up with the most innovative companies that operate an industry with low margins and large turnover seem to be a good starting point.
This case study will show how Big Data and predictive analytics can facilitate process innovation in retail. The case study gives details on the automation of the replenishment process for a major retailer.
Data Science – Learning from the Past
We make use of assisted learning algorithms to calculate precise and robust predictions of the future. Our unique approach allows us to get full probability densities as predictions for single events (like sales for the combination of product, location, and date). This approach allows produces a ton of information about the uncertainty of events and enables us to take a whole new approach for data-driven optimizations regarding balancing availability and write-offs.
Data, Data, everywhere – but is it enough and of sufficient quality?
Getting the project off the ground requires getting your hands on the data – the earlier the better. All customers we have met so far are convinced they have sufficient data of superior quality. The reality looks a bit different. While companies do collect and store data, we often find they do not have the data at hands you need for a specific scenario. This part will cover the typical patterns where data is of poor quality (for predictive approaches) or data is missing (and nobody noticed so far).
Brain- and System-Surgery – Making a company data-driven
Retailers have to make Millions of decisions each day: which product should be delivered into which store at what day. Replenishment – a task that is impossible to complete manually. Basic predictive approaches have long been applied to replenishment. But these approaches do not allow automating the task. Advanced predictive approaches allow for full automation. A close-loop system can be created, leaving the existing IT landscape intact by adding a predictive component. Millions to Billions of sales predictions are calculated in a fixed time window, resulting in an astonishingly small number of orders.
But having a technology that works is not sufficient to convince retailers. Even the most innovative companies want proof that predictive analytics work. In a world full of statistic processes and random events it is one of the most challenging tasks to overcome our human nature. We are constantly on the search for causal relationships and have a hard time to accept a world full of correlations.
Conclusion
Recap on the ingredients of success

  • scenario, customer, industry
  • precise and robust predictions
  • data quantity and quatlity
  • change: technology and humans
Photo of Jan Karstens

Jan Karstens

Blue Yonder

Since joining Blue Yonder in 2011, Jan has taken the role as Head of Development and CTO. Blue Yonder develops and applies state-of-the-art predictive analytics to real-world business problems.

Previously, Jan worked in different positions in the software industry.

Photo of Johanna Fleckner

Johanna Fleckner

Blue Yonder

Johanna is a data scientist and project manager at Blue Yonder, specialized in demand forecasting in the retail industry. She has worked on projects with major retailers. She holds a PhD in high energy physics.

Previously, Johanna worked for the European Organization for Nuclear Research in Switzerland (CERN) where she gained her deep experience with big data, predictive analytics, and machine learning.

Sponsors

Sponsorship Opportunities

For information on exhibition and sponsorship opportunities at the conference, contact Susan Stewart at sstewart@oreilly.com

Media Partner Opportunities

For information on trade opportunities with O'Reilly conferences contact Kathy Yu at mediapartners
@oreilly.com

Press and Media

For media-related inquiries, contact Maureen Jennings at maureen@oreilly.com

Contact Us

View a complete list of Strata contacts