Szepannek, Gero and Aschenbruck, Rabea - Archives of Data Science, Series A

Article Details

Title Predicting eBay Prices: Selecting and Interpreting Machine Learning Models – Results of the AG DANK 2018 Data Science Competition
Authors Szepannek, Gero and Aschenbruck, Rabea
Year 2020
Volume 7(1)
Abstract The annual meeting of the work group on data analysis and numeric classification (DANK) took place at Stralsund University of Applied Sciences, Germany on October 26𝑡h and 27𝑡h, 2018 with a focus theme on interpretable machine learning. Traditionally, the conference is accompanied by a data science competition where the participants are invited to analyze one or several data sets and compare and discuss their solutions. In 2018, the task was to predict end prices of eBay auctions. The paper describes the task as well as a discussion of the results as provided by the conference participants. These cover aspects of preprocessing, comparison of different models, task specific hyperparameter tuning as well as the interpretation of the resulting models and the relevance of additional text information.