Analisis Sentimen Perusahaan Listrik Negara Cabang Ambon Menggunakan Metode Support Vector Machine dan Naive Bayes Classifier
Twitter is one of the most popular online networking service that accessed by the community / citizen. The account @ambonlima which has proven their credibility, take the opportunity to provide the information about the electricity in Ambon Island, Maluku. Power outage in Ambon is become the issue l...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Politeknik Harapan Bersama Tegal
2018-10-01
|
| Series: | Jurnal Informatika: Jurnal Pengembangan IT |
| Online Access: | http://ejournal.poltektegal.ac.id/index.php/informatika/article/view/977 |
| Summary: | Twitter is one of the most popular online networking service that accessed by the community / citizen. The account @ambonlima which has proven their credibility, take the opportunity to provide the information about the electricity in Ambon Island, Maluku. Power outage in Ambon is become the issue lately and the community conveyed via tweets addressed to @ambonlima accounts such as complaints, criticisms or supports. The opinion is textual data which can be extracted to know the sentiment of society to the performance of PT. PLN Ambon. The purpose of this research is to to found out the sentimental level of the society about the electrical condition in Ambon by using sentiment analysis method. There are two classification method used in this research, Naive Bayes Classifier (NBC) dan Support Vector Machine (SVM). In this case, the researcher will compare both method to understand which method have better accuracy. The NBC classification results using 2 fold in validation process showed a better accuracy than other fold value, which is 67.2%. Positive sentiment obtained 67%, neutral sentiment 19% and negative sentiment 14%. Meanwhile, the SVM classification method also showed a better accuracy using 2 fold. Positive sentiment derived 24%, neutral sentiment 29%, and negative sentiment 47%. This study shows the average level of accuracy of SVM classification method is better than the NBC method, which is 76.42%. The presence of negative sentiment that is not more than 50% indicates the influence of account @ambonlima which is able to afford the electrical problems to the public in real time. |
|---|---|
| ISSN: | 2477-5126 2548-9356 |