Analisis Sentimen Pengguna Sistem Pay Later Menggunakan Support Vector Machine Metode Pembobotan Lexicon

Pay Later System User Sentiment Analysis Using Support Vector Machine Lexicon Weighting Method

Layanan pay later sangat mudah dengan cepat populer di masyarakat, hal ini disebabkan karena fitur ini cenderung mudah digunakan dan populer karena tertanam pada e-commerce. Banyaknya penyedia layanan pay later menyebabkan diperlukannya pemilahan penyedia paylater mana yang akan digunakan berdasarkan berbagai pertimbangan termasuk review dari pengguna lain. Di sisi lain, review pengguna lain dapat diperoleh dari Twitter. Data dari Twitter menunjukkan terdapat ribuan tweets pada tahun 2021 hingga 2022 berisikan opini masyarakat terkait penggunaan pay later. Tweets tersebut membahas tentang keunggulan, keluhan, dan ulasan dari penggunaan pay later. Namun, beberapa keunggulan, keluhan, dan ulasan tersebut banyak yang bersifat abstrak sehingga masih belum optimal pemanfaatannya. Penelitian ini bertujuan untuk mengklasifikasikan data tweet yang berkaitan dengan pay later menggunakan algoritma Support Vector Machine (SVM). Dari penelitian ini telah berhasil dibangun model klasifikasi SVM untuk kasus sentimen Shopee Paylater dan Gopay Later. Pada pemodelan sentimen Shopee Paylater diperoleh bahwa model telah dapat memprediksi kelas data uji dengan akurasi 89.74%. Pada pemodelan sentimen Gopay Later diperoleh bahwa model telah dapat memprediksi kelas data uji dengan akurasi 90.27%.

Kata kunci: pay later, SVM, klasifikasi, analisis sentimen

Pay later services are very easy and popular in the community because it tends to be easy to use and is embedded in e-commerce. Due to the vast number of pay later service providers, it is necessary to choose which pay later provider to use based on a variety of factors, including evaluations from other users. On the other hand, other user reviews can be obtained from Twitter. Twitter data shows that there are thousands of tweets from 2021 to 2022 containing public opinion regarding the use of pay later. The tweets discuss the advantages, complaints, and reviews of using pay later. However, many of these advantages, complaints, and reviews are abstract so that their utilization is still not optimal. This study aims to classify tweet data related to pay later using the Support Vector Machine (SVM) algorithm. From this research, the SVM classification model has been successfully built for the sentiment case of Shopee Paylater and Gopay Later. In the Shopee Paylater sentiment modeling, it was found that the model was able to predict the test data class with an accuracy of 89.74%. In Gopay Later sentiment modeling, it was found that the model was able to predict the class of test data with an accuracy of 90.27%.

Keywords: pay later, SVM, classification, sentiment analysis