Avoid out of stocks and achieve lower inventory levels by exploiting big data with machine learning technology

Argusi has been selected to take part in the European Data Incubator (EDI), a 3-year project that offers around 100 startups and small companies the chance to solve data challenges set by major European corporates like Volkswagen Navarra, RACC motoring club and the multinational Sonae.

We are taking part in EDI to solve the challenge set by Sonae on Supply Chain Optimization. Sonae Center Serviços II, S.A. is a leader in the Portuguese retail market owning hypermarkets & convenience supermarkets, electronics stores, fashion stores, cafeteria and restaurants, bookshops & stationery, health & wellbeing stores.

Our challenge

In the EDI challenge we are exploring new methods to handle the increasingly big data sets we receive in our projects, that can be tedious and slow to work with and do not fit standard software such as MS Access or Excel. And even more exciting: we are looking into the opportunities that big data can bring to supply chain optimization. Specifically, the application of datamining and machine learning technology.

Our Approach

For the Sonae challenge we have worked out a three step plan to reduce safety stock and waste at the retailer, while increasing service levels:

1. Improve forecast on SKU level

By applying data mining technology we search for correlations in the internal big data (such as POS data, promotions, product characteristics, location information) and external big data (such as weather data, holidays, demographic and economic data). We apply machine learning technology to improve the forecast on individual SKU level.

2. Segmentation

Next we cluster products into relevant segments based on their demand pattern, combined with a wide range of factors such as: product value, perishability, assortment, promotions, suppliers, lead times, MOQ, seasonality. We do this by applying machine learning clustering technology, and go far beyond the traditional ABC-XYZ segmentation.

3. Taylormade inventory strategy

Based on the segmentation we determine the best inventory strategy for each product segment. The parameter settings will then be optimized on a SKU level. Then the proposed inventory strategy is validated in a simulation model.

The result will be:

  • Improvement of the ordering process: ordering point and quantity
  • Reduce safety stocks
  • Reduce waste
  • Increase service levels


Focus Group

We are supported in this challenge by a focus group, consisting of experts from (retail) companies and academics. They provide us with feedback on topics such as: business questions, methodology, interfacing, data availability etc. and they help us develop a successful business model.

If you are interested in joining the focus group, please contact us!

What do we ask of participants in the focus group?

  • 1-on-1 interview with one us at the start of the project
  • Workshop with all focus group participants, where our first results will be presented
  • An (skype) interview at the end of the project to evaluate and discuss next steps.


Datathon in Berlin

On 17 and 18 October Menno and Marlies travel to Berlin to represent Argusi in the first gathering of the 30 selected startups. Here we will participate in a datathon, and meet face to face with the companies that supplied the challenges. We will pitch our approach and a jury will decide which 16 startups will pass to the next phase of the project.

To read more about EDI, please visit the website: www.edincubator.eu

Do you have a question or would you like to receive more information?


Send us a message.