News

Game on: Otto Group hopes a little rivalry will help solve its product categorization woes

March 20, 2015 02:52 PM
http://www.internetretailer.com/static/uploads/thumbs/product_category_jpg_78x78_crop_q95.jpg

Techies, listen up. You could win $5,000 for your talents.

Germany-based Otto Group is a huge e-commerce company. As it has grown it’s become difficult for the company to accurately track and classify products across its many e-commerce entities, including Crate & Barrel (USA), Otto.de (Germany) and 3 Suisses (France). So it’s calling on the public for help.

Otto, No. 2 in the Internet Retailer 2014 Europe 500 with 6 billion euros in web sales ($6.48 billion), on Tuesday launched a competition on data science contest site Kaggle.com in which it asks participants to develop an open-source predictive model that can distinguish between Otto’s main product categories. Otto has provided a data set with 93 features for more than 200,000 products for contestants to work with. The competition closes May 18. First place winner takes home $5,000, second place $3,000 and third place $2,000. Open source means anyone can see and modify the source code, unlike proprietary software such as Microsoft Office.

“We are selling millions of products worldwide every day, with several thousand products being added to our product line,” Otto says on Kaggle.com. “A consistent analysis of the performance of our products is crucial. However, due to our diverse global infrastructure, many identical products get classified differently. Therefore, the quality of our product analysis depends heavily on the ability to accurately cluster similar products. The better the classification, the more insights we can generate about our product range.”

Submissions are judged using a mathematical formula, which is explained on the site, but essentially entries are measured by how close each model’s predicted categories are to the actual product categories. Players must submit their entries via a comma separated value (CSV) file. As of 11:30 a.m. Central time the competition had attracted 579 teams, 594 players and 2,092 entries.

In October 2006, streaming movie and TV show subscription service Netflix took a similar approach to garner help developing a better recommendation model by improving the accuracy of predictions about how much a viewer is going to enjoy a movie based on her movie preferences.

After almost three years and submissions by more than 40,000 teams from 186 countries, Netflix awarded the much larger $1 million Netflix Prize to a team of engineers, statisticians and researchers who called themselves “BellKor’s Pragmatic Chaos” and achieved the competition’s goal of a 10% improvement over the accuracy of the Netflix movie recommendation system.

Netflix is No. 7 in the Internet Retailer 2104 Top 500 Guide with $4.375 billion in online sales in 2014.

 

 

Back

 

Top Solution Providers

close