Show simple item record

dc.contributor.advisorWigena, Aji Hamim
dc.contributor.advisorAnisa, Rahma
dc.contributor.authorWahyuningsih, Fetri
dc.date.accessioned2022-10-26T00:20:52Z
dc.date.available2022-10-26T00:20:52Z
dc.date.issued2022
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/115075
dc.description.abstractFlip merupakan start-up kreasi pemuda Indonesia yang memiliki popularitas dan rating tinggi pada situs Google Play Store. Hal tersebut membuat tuntutan terhadap mutu pelayanan aplikasi juga semakin meningkat, sehingga muncul beragam komentar di kolom ulasan. Analisis sentimen merupakan salah satu cara untuk mengetahui polaritas opini yang memungkinkan klasifikasi ulasan ke dalam kelas bersentimen positif, netral, atau negatif. Penelitian ini bertujuan untuk mengidentifikasi perbedaan performa algoritma Support Vector Machine (SVM) menggunakan kernel linear, kernel polynomial, dan kernel Radial Basis Function (RBF) pada tiga fitur n-gram (unigram, bigram, dan trigram) serta melakukan analisis sentimen terhadap ulasan para pengguna aplikasi Flip. Algoritma Multi-Class SVM One Against All (OAA) diterapkan karena terdapat tiga kelas sentimen. Selain itu, algoritma Synthetic Minority Oversampling Technique (SMOTE) diterapkan untuk mengatasi ketidakseimbangan pada data. Pemodelan SVM menggunakan kernel RBF pada fitur unigram menghasilkan performa terbaik dalam mengklasifikasikan sentimen dengan nilai akurasi dan macro average F1-score yang didapat sebesar 89% dan 70%. Secara keseluruhan, kernel linear dan RBF pada ketiga fitur n-gram memberikan hasil klasifikasi terbaik. Sebaliknya, kernel polynomial memberikan hasil klasifikasi terendah. Fitur unigram memberikan hasil klasifikasi terbaik dan fitur trigram memberikan hasil klasifikasi terendah untuk ketiga kernel.id
dc.description.abstractFlip is a start-up created by Indonesian youth that has high popularity and ratings on the Google Play Store site. This makes the demands on the quality of application services also increase, thus various comments appear in the review column. Sentiment analysis is one way to find out the polarity of opinions that allows the classification of reviews into classes that have positive, neutral or negative sentiments. This study aims to identify differences in the performance of the Support Vector Machine (SVM) algorithm using a linear kernel, a polynomial kernel, and a Radial Basis Function (RBF) kernel on three n-gram features (unigram, bigram, and trigram) and to conduct sentiment analysis on the reviews of Flip application users. The Multi-Class SVM One Against All (OAA) algorithm is applied because there are three sentiment classes. In addition, the Synthetic Minority Oversampling Technique (SMOTE) algorithm was applied to overcome the data imbalance. SVM modeling using the RBF kernel on the unigram feature produces the best performance in classifying sentiment with an accuracy value and a macro average F1-score obtained at 89% and 70%. Overall, the linear kernel and RBF kernel on the three n-gram features give the best classification results. On the other hand, the polynomial kernel gives the lowest classification result. The unigram feature gives the best classification result and the trigram feature gives the lowest classification result for the three kernels.id
dc.language.isoidid
dc.publisherIPB Universityid
dc.titleImplementasi Algoritma Support Vector Machine untuk Analisis Sentimen Ulasan Aplikasi Flip pada Situs Google Play Storeid
dc.title.alternativeImplementation of Support Vector Machine Algorithm for Sentiment Analysis of Flip Application Reviews on Google Play Store Siteid
dc.typeUndergraduate Thesisid
dc.subject.keywordflipid
dc.subject.keywordkernelid
dc.subject.keywordn-gramid
dc.subject.keywordsentiment analysisid
dc.subject.keywordsupport vector machineid


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record