Peningkatan Akurasi Algoritma Support Vector Machine (Svm) Dengan Particle Swarm Optimization (Pso) Pada Analisis Sentimen Review Hotel Di Tasikmalaya (Studi Kasus: Tiket.Com Dan Agoda)

Fajrin, Septiana Nugraha (2022) Peningkatan Akurasi Algoritma Support Vector Machine (Svm) Dengan Particle Swarm Optimization (Pso) Pada Analisis Sentimen Review Hotel Di Tasikmalaya (Studi Kasus: Tiket.Com Dan Agoda). Undergraduate Thesis thesis, Institut Telkom Telkom Purwokerto.

	Text Cover_17102032.pdf Download (3MB)
	Text Abstract_17102032.pdf Download (188kB)
	Text Abstrak_17102032.pdf Download (190kB)
	Text BAB I_17102032.pdf Download (250kB)
	Text BAB II_17102032.pdf Download (717kB)
	Text BAB III_17102032.pdf Download (567kB)
	Text BAB IV_17102032.pdf Restricted to Registered users only Download (813kB) | Request a copy
	Text BAB V_17102032.pdf Download (189kB)
	Text Daftar Pustaka_17102032.pdf Download (320kB)
	Text Lampiran_17102032.pdf Restricted to Registered users only Download (373kB) | Request a copy

Abstract

The use of technology, especially in hotel bookings reported from several sites in Indonesia from 2018 to 2021, can be seen on the platforms that are often used by users, namely Tiket.com and Agoda. The data used is 3,295 hotel customer review data in Tasikmalaya on Tiket.com and Agoda. The problem in this study is that there is no sentiment analysis model to determine the increase in the accuracy of hotel reviews in Tasikmalaya on Tiket.com and Agoda. This research has a goal, namely, to find an increase in the accuracy and performance of the customer review model to make it easier for the hotel to present it in the form of higher quality data and information. The method that will be used is the Support Vector Machine (SVM) classification with its advantages, namely minimizing errors in the training-set or can be called Structural Risk Minimization (SRM) and realizing the formation of a hyperplane by maximizing margins and Particle Swarm Optimization (PSO) to improve performance. classification model. The tools used are Python. The study used K-Fold Cross Validation with values of k=5 and k=10. K-Fold Cross Validation with a value of k = 5 means that the data will be randomized five times and the data will be randomized using 80% training data and 20% test data. K Fold Cross Validation with a value of k = 10 means that the data will be randomized ten times and the data will be randomized using 90% training data and 10% test data. The results of the k-Fold Cross Validation study using a linear kernel with k=10 resulted in the greatest accuracy value with an accuracy value of 81.05% and increasing accuracy by 1.05%. The conclusion from the research of Support Vector Machine and Particle Swarm Optimization using a linear kernel produces the highest level of accuracy from the Radial Basis Function (RBF) kernel and the Polynomial kernel. Keywords: Support Vector Machine (SVM), Structural Risk Minimization (SRM), PSO, TF-IDF, Confusion Matrix

Actions (login required)

View Item