AFFM: Auto feature engineering in field-aware factorization machines for predictive analytics
Lars Ropeid Selsaas, Bikash Agrawal, Chunming Rong, Thomasz Wiktorski · 2015
User identification and prediction is one typicalproblem with the cross-device connection. User identification isuseful for the recommendation engine, online advertising, anduser experiences. Extreme sparse and large-scale data makeuser identification a challenging problem. To achieve betterperformance and accuracy for identification a better model withshort turnaround time, and able to handle extremely sparse andlarge-scale data is the key. In this paper, we proposed a novelefficient machine learning approach to deal with such problem. We have adapted Field-aware Factorization Machine's approachusing auto feature engineering techniques. Our model has thecapacity to handle multiple features within the same field. Themodel provides an efficient way to handle the fields in the matrix. It counts the unique fields in the matrix and divides both thematrix with that value, which provide an efficient and scalabletechnique in term of time complexity. The accuracy of the modelis 0.864845, when tested with Drawbridge datasets released in thecontext of the ICDM 2015 Cross-Device Connections Challenge.