Evaluasi Dan Prediksi Hasil Pertandingan Dota 2 Menggunakan Algoritma Terbaik Dari Random Forest Atau XGBoost

Fandi, Rico Satria (2024) Evaluasi Dan Prediksi Hasil Pertandingan Dota 2 Menggunakan Algoritma Terbaik Dari Random Forest Atau XGBoost. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 19082010104.-cover.pdf Download (2MB)
	Text (BAB I) 19082010104.-bab1.pdf Download (204kB)
	Text (BAB II) 19082010104.-bab2.pdf Restricted to Repository staff only until 1 July 2026. Download (464kB) \| Request a copy
	Text (BAB III) 19082010104.-bab3.pdf Restricted to Repository staff only until 1 July 2026. Download (245kB) \| Request a copy
	Text (BAB IV) 19082010104.-bab4.pdf Restricted to Repository staff only until 1 July 2026. Download (1MB) \| Request a copy
	Text (BAB V) 19082010104.-bab5.pdf Download (193kB)
	Text (Daftar Pustaka) 19082010104.-daftarputaka.pdf Download (159kB)

Abstract

The development of electronic sports competitions, or esports, has garnered the interest of many people in predicting match outcomes. Dota 2, as one of the most popular MOBA (Multiplayer Online Battle Arena) games, is the focus of this thesis to evaluate factors such as hero selection and item choice in determining victory or defeat. This thesis aims to evaluate and predict Dota 2 match outcomes using Random Forest and XGBoost algorithms. A total of 100,000 match data from patches 7.35—7.35d were collected through the OpenDota API for free. Modeling was conducted using both algorithms with data split scenarios of 80:20, 75:25, and 70:30. Model evaluation was performed using the Confusion Matrix and Area Under The Receiver Operating Characteristic (AUROC). The research results show that the 80:20 scenario with the XGBoost algorithm provided the best performance with an accuracy of 52% on the Confusion Matrix and an AUROC of 51.49%, indicating an optimal balance between accuracy, precision, recall, and f1-score for both Dire Win and Radiant Win classes. The selected XGBoost model was implemented in a Flask framework as the backend to develop a web-based Dota 2 match prediction application. This implementation includes the development of an interface using HTML and Tailwind CSS, ensuring user-friendly functionality. Validation testing on 10,000 new match data showed an accuracy rate of 50.62%. Although XGBoost showed better results compared to Random Forest, this thesis concludes that the model's accuracy level still needs improvement to achieve more accurate predictions.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Arifiyanti, Amalia Anjani	NIDN0712089201	amalia_anjani.fik@upnjatim.ac.id
Thesis advisor	Wati, Seftin Fitri Ana	NIDN0020039104	seftin.fitri.si@upnjatim.ac.id

Subjects:

T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.882 Internet

Divisions:

Faculty of Computer Science > Departemen of Information Systems

Depositing User:

Rico Satria Fandi Fandi

Date Deposited:

01 Jul 2024 07:18

Last Modified:

01 Jul 2024 07:18

URI:

https://repository.upnjatim.ac.id/id/eprint/25373