DATA MINING CUP 2017: How does dynamic pricing affect sales performance?

The announcement of the task yesterday marked the start of the DATA MINING CUP (DMC for short) 2017. The focus of this year’s international student competition for intelligent data analytics is on dynamic pricing and sales prediction.

Dynamic pricing in retail is still a hot topic. Personalized pricing in particular is attracting the attention of consumer protection groups. Why? In certain sectors, like when booking flights online for example, there are practices that are difficult for customers to understand. Obtaining different prices depending on what end device is used or which website the customer visits to access the deal does nothing to instill confidence in the customer. But dynamic pricing encompasses much more: first and foremost the automated adjustment of prices in keeping with the market. Large online shops featuring a wide range of products in particular rely on the automatic adjustment of item prices. In this case, the focus is not on the personalized price for that customer but rather price optimization on the product level. Thanks to regular price adjustment a fair market price is determined for each product. Depending on the market situation, prices get either more expensive or cheaper for the customer. Retailer parameters such as turnover and gross profit are maximized in the process. Dynamic pricing is thus based on demand, supply and the environment and also offers advantages for end customers.

Developing a suitable prediction model: Observe the interplay between product attributes and prices

Participants in this year’s DATA MINING CUP are closely examining the relationships between item prices and product attributes in the context of dynamic pricing. To do this, historical anonymized transaction data including prices, sales figures and different product attributes of an actual online pharmacy are available. Information regarding customer behavior is also known: this includes clicks on products, the creation of shopping baskets and ordering procedures. The online shop uses dynamic pricing in the form of daily automated price adjustments. However it is not only the price but also the interplay between different product attributes (e.g. product group) that impact sales performance.

The goal for the student teams is to develop a model that predicts the sales figures for the following month. The team whose prediction comes closest to the actual sales performance wins the DATA MINING CUP 2017.

Student teams from all over the world put their data mining expertise to the test

The DATA MINING CUP was started by prudsys AG in 2000. Every year, numerous student teams from all over the world take part in the competition for intelligent data analytics. Past DMC competitions have highlighted a number of topics including return predictions in online retail, forecasts for the participation period of customers in the lottery industry, predictions for coupon redemption rates and payment problems and even predictions for purchases when using dynamic pricing. What is special about the DMC tasks is that they are very practical: Real companies ask practical questions and provide anonymized data for the students to use. We see the DATA MINING CUP as a bridge between theory and practice: student teams get insight into the real market economy. Resonance with the competition is consistently high: This year there are already 135 teams from 106 educational institutions in 30 countries registered. Teams have a total of six weeks to submit their prediction models regarding the sales performance of an online pharmacy. The deadline is May 17 at 14:00 CEST (2 o’clock p.m. UTC+2, or CEST). We recommend that all teams submit their solutions in a timely manner, preferably not five minutes before the deadline. Experience has shown that mistakes can even be made on the home stretch (e.g. incorrectly naming the solution file, wrong return email address, not taking into account the time difference). It is always a good idea to have some buffer time up your sleeve.

Winners are recognized at the prudsys personalization summit on June 27/28 in Berlin

Every year our prudsys DMC team, consisting of colleagues from the research and marketing departments, waits anxiously for the teams to submit their solutions and to have them evaluated. Once the solutions have been ranked, we send out invitations to the ten best teams and keep the standings under wraps until the prudsys personalization summit on June 27/28, 2017 at the nhow Hotel in Berlin. The best teams present their solutions at our annual conference for omnichannel retail. The winning teams are announced during the evening event at the prudsys personalization summit on June 27, 2017. We are already excited to find out who will take home this year’s trophy and hope that all of the teams have fun solving the problem.