ABSTRACT
In this paper, we describe the winning approach for the RecSys Challenge 2015. Our key points are (1) two-stage classification, (2) massive usage of categorical features, (3) strong classifiers built by gradient boosting and (4) threshold optimization based directly on the competition score. We describe our approach and discuss how it can be used to build scalable personalization systems.
- D. Ben-Shimon, A. Tsikinovsky, M. Friedman, B. Shapira, L. Rokach, and J. Hoerle. Recsys challenge 2015 and the yoochoose dataset. In Proceedings of the 9th ACM conference on Recommender systems. ACM, September 2015. Google ScholarDigital Library
- J. Friedman. Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29:1189--1232, 2000.Google Scholar
- A. Gulin. Matrixnet. Technical report, 2012. http://www.slideshare.net/yandex/matrixnet.Google Scholar
- R. Kohavi and C.-H. Li. Oblivious decision trees graphs and top down pruning. In Proceedings of the 14th International Joint Conference on Artificial Intelligence - Volume 2, pages 1071--1077, 1995. Google Scholar
- I. Tolstikhin and Y. Seldin. Pac-bayes-empirical-bernstein inequality. In C. B. et. al., editor, Advances in Neural Information Processing Systems 26, pages 109--117. 2013.Google Scholar
- X. Yi, L. Hong, E. Zhong, N. N. Liu, and S. Rajan. Beyond clicks: dwell time for personalization. In Proceedings of the 8th ACM Conference on Recommender systems, pages 113--120. ACM, 2014. Google Scholar
Index Terms
- RecSys Challenge 2015: ensemble learning with categorical features
Recommendations
RecSys Challenge 2015 and the YOOCHOOSE Dataset
RecSys '15: Proceedings of the 9th ACM Conference on Recommender SystemsThe 2015 ACM Recommender Systems Challenge offered the opportunity to work on a large-scale e-commerce dataset from a big retailer in Europe which is accepting recommender system as a service from YOOCHOOSE. Participants tackled the problem of ...
In-House Solution for the RecSys Challenge 2015
RecSys '15 Challenge: Proceedings of the 2015 International ACM Recommender Systems ChallengeRecSys Challenge 2015 is about predicting the items a user will buy in a given click session. We describe the in-house solution to the challenge as guided by the YOOCHOOSE team. The presented solution achieved 14th place in the challenge's final ...
RecSys Challenge 2014: an ensemble of binary classifiers and matrix factorization
RecSysChallenge '14: Proceedings of the 2014 Recommender Systems ChallengeIn this paper we give our solution to the RecSys Challenge 2014. In our ensemble we use (1) a mix of binary classification methods for predicting nonzero engagement, including logistic regression and SVM; (2) regression methods for directly predicting ...
Comments