Modeli za predviđanje ponašanja korisnika digitalne igre i trajanja aktivnosti igrača

master's thesis

• Title (english): Models for Prediction of Player Behavior and the Duration of Player Activities in Digital Games
• Author: Sanja Jerković
• Mentor: Lea Skorin-Kapov (mentor)
• Committee member: Lea Skorin-Kapov (predsjednik povjerenstva)
• Committee member: Mirko Sužnjević (član povjerenstva)
• Committee member: Maja Matijašević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Digitalne igre na free-to-play tržišnom modelu sve više postaju popularne. Ponašanje korisnika kroz igru može biti smjernica tome hoće li korisnik odustati od igre. Analizom velikih skupova podataka promatra se korisnikovo ponašanje u svrhu poboljšanja igre i shvaćanja budućih poteza unutar nje. Definiranje skupa podataka za analizu te izbor tipa analize podataka ovisi o vrsti i mehanici igre.
U ovom radu analizirani su stvarni podaci dobiveni od tvrtke koja nudi free-to-play mobilnu igru. Od prije je bilo poznato kako je analiza dugotrajnih korisnika vrlo bitna u svrhu unaprjeđenja ostanka i interakcija korisnika. Isto tako, analizirana free-to-play mobilna igra sadrži dio koji je dokazano najigraniji te je pokazana potreba za razvojem modela koji predviđa buduće ponašanje korisnika i odluke o odustajanju od igre ovisno o upravo tom dijelu igre. Konkretno, korištene su opisna i prediktivna analiza. Kroz analizu sličnih radova i potražnje na tržištu mobilnih free-to-play igara, objašnjena je potreba za analizom podataka o korisnicima. Isto tako, prikazani su korišteni algoritmi strojnog učenja s raznim parametrima. U radu je definiran skup dugotrajnih korisnika i opisano njihovo ponašanje kroz igru. Izgrađen je model za predviđanje korisnikovog odustajanja od igre te prikazana analiza dobivenih rezultata.

Abstract (english)

Digital games on the free-to-play market model are becoming increasingly popular. User behavior throughout the game can be an indicator as to whether the user will give up on the game. The analysis of large datasets observes the user's behavior in order to improve the game and understand future user moves within the game. Defining a user behavior dataset for analysis, and choosing the type of data analysis depends, on the type and mechanics of the game.
This thesis is based on data obtained from a company offering a free-to-play mobile game. It has long been known that long-term user analysis is very important for the purpose of improving user retention and interaction. Also, the analyzed free-to-play mobile game contains a part of the game (in this case a given book read by players) that has been proven to be the most “played” part of the game. Thus, a model was developed in the scope of the thesis that predicts future user’s behavior and decision to give up on the game depending on that part of the game. In particular, the thesis is focused on descriptive and predictive analysis. Through an analysis of related work and market demand for mobile free-to-play games, the need for user data analysis was explained. Furthermore, the thesis presents the used machine learning algorithms and reports on a parameter analysis. The thesis defines a set of long-term users and describes their behavior. Using machine learning algorithms, this thesis presents models for predicting the user’s quitting of the game and analyzes given results.

• Keywords: free-to-play mobilne igre; strojno učenje; analiza velikih skupova podataka; predviđanje odustajanja
• Keywords (english): free-to-play mobile games; machine learning; big data analysis; churn prediction
• Language: croatian
• URN:NBN: urn:nbn:hr:168:031582
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 15:10:55