Destination prediction is crucial for many location-based applications, such as sightseeing recommendations, advertising, and governance. The accurate prediction can help to solve travel demand estimation, traffic congestion, and transit planning. Accordingly, this topic was related to one of the BigData Cup competitions at the 2022 IEEE International Conference on Big Data.This paper presents a pipeline with two steps: candidate retrieval to generate the training and testing set and model training by using the extreme gradient boosting model (LightGBM [1]) with features engineering from trip data and parameter tuning. The solution achieved first place with a categorization accuracy score of 0.43967 in the final result.

Prediction of who is going to win in video games is an important machine learning application [1] [2]. Accordingly, one of the BigData Cup competitions at the 2021 IEEE International Conference on Big Data [3]1was related to this topic. The goal of this challenge is to find a way to predict victories of a video game named Tactical Troops: Anthracite Shift [4]2by game logs. In this kind of challenge, ensemble models are often to be used to get a more accurate result. In this paper, we use a single extreme gradient boosting model (XGBoost [5]) with a wide variety of features from game logs and metadata. The proposed model and features achieved 3rd place with an AUC score of 0.8978 in the final result.