Learning to rank for top mobile games prediction
Alfian Ramadhan, Masayu Leylia Khodra · 2014
Learning to rank as one of machine learning techniques has been widely used in document categorization, information retrieval, text processing, product rating, and other ranking problem domains. Ranking task has become an interesting research in terms of big data and analytics in data mining. Meanwhile, top mobile games prediction is one of the popular topics in terms of increasing games excitement in mobile devices today. This paper introduces an approach for top mobile games prediction using learning to rank. This paper proposes a prediction system which is called MGPrediction. Experiments have been conducted by performing our prediction system and executing various algorithms to our ranking dataset of mobile games using learning to rank in RankLib. Preliminary results present that our system has been successfully applied and we have discovered that MART gives the highest accuracy among others in training model and ranking prediction samples. Thus, MART is the most considered algorithm towards top mobile games predictions problem.