DATA MINING CUP 2018

193 Teams from 148 universities in 47 countries

Scenario

Dynamic pricing strategies are increasingly common, particularly in online business. For the purposes of simplification, long product lifecycles and an inexhaustible stock are often assumed. Conversely, in the area of fashion it seems expedient to take into account both the product age and the stock. The product lifecycles are mostly very short in this area. In this context it is very important for the retailer to have sold out an item at a particular time if possible, as the subsequent item then appears. It is equally important to know when an item will be sold out, to reorder on cue if the item has not yet reached the end of its life.

A sporting goods retailer uses dynamic prices to control when items sell out in their online shop. A good prediction of when items will be sold out is necessary in order to make the most expedient price adjustments. The time a product is sold out depends not just on its price, but also on other product attributes, such as brand, size and product group.

Task

The task for the participating teams is to use the sales data from a period of four months to develop a prediction model, which can be used to predict the products time of sell out in the following month. The aim is to predict as accurately as possible the precise day when items will sell out.

Downloads

Please enter the password from the team leader mailing in order to get to download the zip archive (1,44 MB) with task and data files.

Task
Solution

Winners

First Place:

Team TH_Zuerich_1

Winning team of the DATA MINING CUP 2018

Prize: 2,000.00 EUR

Second Place:

Team Uni_Mannheim_2

Second place of the DATA MINING CUP 2018

Prize: 1,000.00 EUR

Third Place:

Team TH_Zuerich_2

Third place of the DATA MINING CUP 2018

Prize: 500.00 EUR