Wara, Shindi Shella May (2019) Analisis Respons Warganet Terhadap Debat Calon Presiden 2019 di Twitter dengan Metode Support Vector Machines. Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211540000101_Undergraduate_Thesis.pdf Download (1MB) | Preview |
Abstract
Debat calon presiden 2019 merupakan salah satu sarana kampanye yang disediakan oleh Komisi Pemilihan Umum (KPU) dimana pasangan calon presiden saling berhadapan untuk menyampaikan visi dan misi serta tujuan yang dibawa untuk masa depan Indonesia. Debat calon presiden 2019 disiarkan langsung secara nasional sehingga masyarakat dapat secara langsung menanggapi dan lebih mengenal calon presidennya. Bagi warganet dapat secara langsung menyampaikan pendapat terhadap debat presiden 2019 pada media sosial, salah satunya melalui Twitter. Tanggapan yang diberikan bisa berupa sentimen positif, negatif, dan netral. Menyikapi keadaan tersebut, dilakukan penelitian tentang analisis klasifikasi sentimen dengan metode Clustered Support Vector Machines. Data yang digunakan dalam penelitian ini diambil dari tweets dari pengguna Twitter Indonesia yang dipublikasikan pasca debat I-V dengan menyebutkan salah satu nama akun twitter pasangan calon presiden Indonesia 2019. Hasil klasifikasi menunjukkan bahwa pasangan calon 1 lebih banyak mendapatkan tweets sentimen positif dan negatif. Penelitian ini membandingkan metode CSVM berdasarkan nilai akurasi dan AUC pada kernel Linear, Polinomial, dan RBF. Hasil terbaik akurasi dan AUC menggunakan Kernel Polinomial pada Pasca Debat I-IV dengan Akurasi sebesar 0,87; 0,85; 0,88; dan 0,87 dan AUC sebesar 0,90; 0,88; 0,90; dan 0,90. Pada pasca debat V, kernel RBF menghasilkan hasil terbaik dengan Akurasi dan AUC masing-masing sebesar 0,87 dan 0,90.
=================================================================================================================================
Debate of president candidates in 2019 is one of campaign endorsed by “Komisi Pemilihan Umum” (KPU) where president candidates are facing each other to convey their vision, mission and the purpose they brought for the future of Indonesia. Debate of president candidates in 2019 was live broadcasted nationally so that people could respond and have an understanding on their president candidates. In social media, people could directly convey their opinions by online towards presiden debate 2019, one of them is via Twitter. Responses given by people through social media could be positive sentiment, negative or neutral. Given these situations, this research about analysis of sentiment using Clustered Support Vector Machines. The data used in this research were obtained from tweets from Indonesian users that publicated on post-debate I-V that mentioned one of president candidates twitter account. The result of classification show that candidate number 1 have more positive and negative sentiment tweets. The purpose of this research is to compare the accuracy and AUC value in Linear, Polinomial and RBF Kernel. Thus, the highest accuracy and AUC was obtained by using Polinomial Kernel on post-debate I-IV with the accuracy of 0,87; 0,85; 0,88; and 0,87 and the AUC value of 0,90; 0,88; 0,90 and 0,90. In post-debate V, RBF Kernel produce the highest accuracy of 0,87 and AUC value of 0,90.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Calon Presiden, Debat, K-Means, Sentimen, SVM |
| Subjects: | Q Science > QA Mathematics Q Science > QA Mathematics > QA278.55 Cluster analysis Q Science > QA Mathematics > QA353.K47 Kernel functions (analysis) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Shindi Shella May Wara |
| Date Deposited: | 14 Mar 2024 07:13 |
| Last Modified: | 14 Mar 2024 07:13 |
| URI: | http://repository.its.ac.id/id/eprint/64282 |
Actions (login required)
 |
View Item |