Seguí Mesquida, SantiGilabert Roca, PereMoreno Blanco, MarcosTorralba Agell, Adrià2022-05-242022-05-242021-07-01https://hdl.handle.net/2445/185952Treballs finals del Màster de Fonaments de Ciència de Dades, Facultat de matemàtiques, Universitat de Barcelona. Curs: 2020-2021. Tutor: Santi Seguí Mesquida i Pere Gilabert Roca[en] Recommendation systems is an interesting and wide field of research and it is present in a huge amount of different areas in our daily life. The RecSys ACM conference is the most important conference in the recommendation area and every year they organise a competition: the RecSys Challenge. The work presented here aims to solve the RecSys 2021 Challenge which consists of giving a probability to two Twitter users that interact. In this project we have worked in the development of a model which uses the power of Gradient Boosting Trees to combine multiple hand-crafted features in an aim to represent the interaction between the users. Our team reached the 14th place in the overall challenge leaderboard and is placed between the 7th and the 9th place in terms of Like overall performance.60 p.application/pdfengcc-by-nc-nd (c) Marcos Moreno Blanco i Adrià Torralba Agell, 2021codi: GPL (c) Marcos Moreno Blanco i Adrià Torralba Agell, 2021http://creativecommons.org/licenses/by-nc-nd/3.0/es/http://www.gnu.org/licenses/gpl-3.0.ca.htmlSistemes d'ajuda a la decisióXarxes socials en líniaTeoria de la prediccióTreballs de fi de màsterDades massivesDecision support systemsOnline social networksPrediction theoryMaster's thesesBig datainfo:eu-repo/semantics/bachelorThesisinfo:eu-repo/semantics/openAccess