-
Notifications
You must be signed in to change notification settings - Fork 0
/
bab5.tex
29 lines (21 loc) · 2.98 KB
/
bab5.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
\chapter{PENUTUP}
\label{sec:chap5_tutup}
\vspace{1ex}
\section*{}
Setelah penerapan metode terhadap masalah yang ingin diselesaikan pada Bab \ref{chap:chap3_metodologi} dan dilakukan pengujian terhadap metode tersebut pada Bab \ref*{chap:chap4_eval} maka didapatlah kesimpulan yang akan dijabarkan pada Sub-bab berikut. Kemudian dilanjutkan dengan saran untuk penelitian kedepannya pada Sub-bab selanjutnya.
\vspace{1ex}
\section{Kesimpulan}
\label{sec:sec4_kesimpulan}
\vspace{1ex}
Pada saat pengerjaan metodologi pada BAB \ref{chap:chap3_metodologi} yang dilanjutkan dengan pengujian yang tertuang pada BAB \ref{chap:chap4_eval}, mengharuskan untuk membagi permainan dengan genre RPG ke dalam dua golongan lagi yaitu \textit{single-character} dan \textit{multi-character}. Hal ini dilakukan demi memudahkan pembuatan atribut \textit{gameplay}, \textit{single-character} biasanya digunakan pada permainan RPG dengan genre WRPG, ARPG, SRPG, dan MMORPG, sedangkan \textit{multi-character} biasanya digunakan pada permainan RPG dengan genre TRPG dan JRPG.
\vspace{1ex}
Digunakannya metode $k$-NN, Distribusi Normal, dan Naive Bayes dalam melakukan proses distribusi atribut \textit{gameplay} dalam permainan RPG menjadikan atribut \textit{gameplay} dari pemain dan musuh menjadi terpola dari level terendah ke level tertinggi untuk pemain, mejadi tersebar dengan merata untuk distribusi musuh yang harus dihadapi oleh pemain disetiap levelnya. Hal semacam itu akan memudahkan pengembang permainan dalam mendesain permainan, terlebih lagi dalam pembuatan permainan dengan level yang tinggi, karakter, dan musuh yang berjumlah banyak.
\vspace{1ex}
Penggunaan \textit{Neural Network} untuk klasifikasi \textit{hero} pada permainan Dota 2 mampu mencapai 64\% \textit{hero} berhasil terklasifikasi. Sedangkan untuk klasifikasi karakter pemain yang dibuat menggunakan metode $k$-NN dan Naive Bayes dengan jumlah 8 karakter mampu mencapai 50\% pada data \textit{testing}, berarti 50\% dari karakter sesuai dengan tipe yang ingin dibuat. Kemudian untuk klasifikasi karakter musuh yang juga dibuat menggunakan metode tersebut dengan jumlah 400 karakter mampu mencapai 42.5\% pada data \textit{testing}, berarti 42.5\% dari karakter sesuai dengan tipe yang ingin dibuat.
\vspace{1ex}
Tinggi dan rendahnya hasil dari klasifikasi sangat bergantung pada pola atribut \textit{gameplay} yang dihasilkan itu sendiri, dari pola atribut \textit{gameplay} tersebut mampu membentuk koerelasi fitur dengan sekumpulan fitur yang berpengaruh terhadap benarnya hasil klasifikasi. Maksud dari berpengaruh disini adalah mampu membentuk pola atribut \textit{gameplay} yang sesuai dengan tipe.
\vspace{1ex}
\section{Saran}
\label{sec:sec4_saran}
\vspace{1ex}
Dalam penelitian kedepannya sangat disarankan untuk meneliti tentang bagaimana mensimulasikan pertaarungan secara otomatis pada permainan genre RPG. Hal tersebut bertujuan guna mencoba mensimulasikan nilai yang dihasilkan dari penelitian ini yang berupa atribut \textit{gameplay}.