Year of Graduation
Purchase Prediction in Streaming Data
Applied Mathematics and Information Science
In 2015, on the site RecsysChallenge 2015 was organized competition for the prediction of sessions that will end with the purchase and items that will be purchased in these sessions based on a file with clicks of users of one of the online stores. This work focuses on considering an additional possible approach to solving this problem. The task is divided into two subtasks of classification. The main attention is paid to such methods of machine learning as XGBoost, Adaboost and Random Forest. As a result, a competitive model was obtained, which is included in the top 6 solutions based on the result of the proposed scoring function from more than 500 suggested solutions.