Optimasi Parameter Sistem Reward pada Ekonomi Gim Menggunakan Algoritma Evolusier

Satriawan, Alridho Tristan (2026) Optimasi Parameter Sistem Reward pada Ekonomi Gim Menggunakan Algoritma Evolusier. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002221100-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (45MB) | Request a copy

Abstract

Permasalahan reward balance dalam desain ekonomi gim merupakan faktor kritis yang mempengaruhi keterlibatan dan keinginan pengalaman bermain. Sistem reward yang tidak seimbang menyebabkan menurunnya minat pemain, perkembangan permainan yang tidak wajar, serta ketidakstabilan ekonomi di dalam gim. Penelitian terdahulu seperti GEEvo telah mengotomatisasi penyeimbangan gim ekonomi menggunakan algoritma evolusioner, namun belum mempertimbangkan variabilitas perilaku pemain. Penelitian ini mengembangkan dua kontribusi utama pada tahap Balancer, yaitu simulasi berbasis agen dengan tiga profil perilaku (agresif, pasif, acak) untuk merepresentasikan dinamika pemain dan fungsi kebugaran obj4 untuk memancarkan jarak dekat nilai sumber daya terhadap target. Evaluasi komparatif pada berbagai graf abstrak menunjukkan bahwa kondisi C, dalam hal ini yang menggunakan sistem simulasi agen dan fungsi fitness obj4 merupakan kondisi terbaik secara signifikan. Pada graf bebas, kondisi C mencapai Balanced % yang lebih tinggi dibandingkan kondisi A sebagai baseline dan kondisi B yang masih menggunakan fungsi obj3. Pada grafik terpenuhi, kondisi C mencapai tingkat Balanced % dan Improved % yang tinggi serta mempercepat konvergensi generasi. Perbandingan kondisi B (agen + obj3), obj4 secara signifikan menurunkan median generasi yang dibutuhkan. Selain itu, dikembangkan modul ekstraksi otomatis konfigurasi ekonomi dari Game Design Document (GDD) menggunakan pencocokan kata kunci. Validasi pada GDD MMORPG dan Trading Game pada berbagai sampel graf menunjukkan keberhasilan yang bervariasi.
================================================================================================================================================
The issue of reward balancing in game economy design is a critical factor that affects player engagement and the sustainability of the gaming experience. An unbalanced reward system leads to decreased player interest, irregular game progression, and economic instability within the game. Previous research, such as GEEvo , has automated game economy balancing using evolutionary algorithms, but has not yet considered player behavior variability. This research develops two main contributions in the Balancer stage, namely an agent-based simulation with three behavioral profiles (aggressive, passive, random) to represent player dynamics, and an obj4 fitness function to evaluate the proximity of resource values to the target. A comparative evaluation on various abstract graphs shows that condition C, which utilizes the agent simulation system and the obj4 fitness function, is significantly the best-performing condition. On free graphs, condition C achieves a higher Balanced % compared to condition A as a baseline and condition B which still uses the obj3 function. On isolated graphs, condition C achieves high levels of Balanced % and Improved %, while accelerating generation convergence. Compared to condition B (agent + obj3), obj4 significantly reduces the median generations required. Additionally, an automatic extraction module for economy configurations from Game Design Documents (GDD) was developed using keyword matching. Validation on MMORPG and Trading Game GDDs across various graph samples demonstrates varying levels of success.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Algoritma Evolusioner, Simulasi Agen, Reward Balancing, Game Economy, Game Design, Evolutionary Algorithm, Agent-based Simulation
Subjects:	Q Science > QA Mathematics Q Science > QA Mathematics > QA9.58 Algorithms
Divisions:	Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Alridho Tristan Satriawan
Date Deposited:	31 Jul 2026 04:05
Last Modified:	31 Jul 2026 04:05
URI:	http://repository.its.ac.id/id/eprint/140649

Actions (login required)

View Item