Monitoring Sentimen Ulasan Masyarakat di Google Maps Terhadap Layanan Terminal Purabaya dengan Diagram Kendali p Berbasis Multilingual Bidirectional Encoder Representations From Transformers (mBERT)

Satriya, Muhammad Noor (2026) Monitoring Sentimen Ulasan Masyarakat di Google Maps Terhadap Layanan Terminal Purabaya dengan Diagram Kendali p Berbasis Multilingual Bidirectional Encoder Representations From Transformers (mBERT). Other thesis, Insttitut Teknologi Sepuluh Nopember.

Text 5003221140-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB) | Request a copy

Abstract

Terminal Purabaya atau sering dikenal sebagai terminal Bungurasih merupakan simpul transportasi vital di Jawa Timur dengan mobilitas penumpang yang sangat tinggi. Namun, volume aktivitas yang besar sering kali memunculkan berbagai keluhan terkait kualitas layanan, seperti keberadaan calo, fasilitas yang kurang memadai, dan kebersihan yang buruk. Pemantauan kepuasan pelanggan secara manual menjadi tidak efisien mengingat besarnya jumlah data ulasan. Penelitian ini bertujuan untuk memonitoring kualitas layanan Terminal Purabaya berdasarkan ulasan pengguna di Google Maps menggunakan pendekatan klasifikasi sentimen murni. Metode yang digunakan adalah Multilingual Bidirectional Encoder Representations from Transformers (mBERT) yang diintegrasikan dengan Statistical Process Control (SPC) berupa diagram kendali p. Data yang diekstraksi merupakan ulasan pada periode 1 Januari 2021 hingga 4 Desember 2025. Berdasarkan hasil pelatihan dan pengujian, model mBERT terbaik menunjukkan performa excellent classification dengan nilai akurasi testing sebesar 90,45% dan Area Under Curve (AUC) sebesar 92,11%. Hasil klasifikasi menunjukkan bahwa ulasan layanan didominasi oleh sentimen positif (72,69%), diikuti sentimen negatif (26,45%), dan sentimen netral (0,86%). Pada tahap pemantauan SPC, analisis diagram kendali p fase I untuk data ulasan hasil prediksi berhasil mencapai keadaan terkendali secara statistik pada iterasi ke-9 dengan batas rata-rata proporsi cacat sebesar 0,245. Pada pemantauan fase II, masih ditemukan titik pengamatan yang berada di luar kendali (out of control). Hasil monitoring dengan diagram kendali p menggunakan data rating dan ulasan pengguna terminal memperlihatkan bahwa masih terdapat beberapa pengamatan yang belum terkendali secara statistik yang berarti pihak Terminal Purabaya masih perlu melakukan evaluasi dan peningkatan kinerja pelayanan yang ada. Adapun jenis kendala tertinggi yang sering dikeluhkan oleh pengguna terminal adalah terkait keamanan pengguna terminal terhadap makelar tiket dan copet di area terminal.
=====================================================================================================================================
Purabaya Terminal, often known as Bungurasih Terminal, is a vital transportation hub in East Java with very high passenger mobility. However, the large volume of activity often gives rise to various complaints regarding service quality, such as the presence of touts, inadequate facilities, and poor hygiene. Manual monitoring of customer satisfaction is inefficient given the large amount of review data. This study aims to monitor the service quality of Purabaya Terminal based on user reviews on Google Maps using a pure sentiment classification approach. The method used is Multilingual Bidirectional Encoder Representations from Transformers (mBERT) integrated with Statistical Process Control (SPC) in the form of a p-control chart. The extracted data consists of reviews from the period of January 1, 2021, to December 4, 2025. Based on the training and testing results, the best mBERT model demonstrated Excellent Classification performance with a testing accuracy of 90,45% and an Area Under Curve (AUC) of 92,11%. The classification results showed that service reviews were dominated by positive sentiment (72,69%), followed by negative sentiment (26,45%), and neutral sentiment (0,86%). In the SPC monitoring stage, the phase I p-control chart analysis for the predicted review data successfully achieved a statistically in-control state at the 9th iteration with an average defect proportion limit of 0,245. In the phase II monitoring, observation points that were out of control were still found. The monitoring results using the p-control chart on rating data and terminal user reviews showed that there were still several observations that were not statistically in control, meaning that the Purabaya Terminal management still needs to evaluate and improve existing service performance. The highest type of constraint frequently complained about by terminal users is related to user security against touts and pickpockets in the terminal area.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Sentimen, SPC, mBERT, Diagram Kendali p, Terminal Purabaya, Sentiment Analysis, SPC, mBERT, p Control Chart, Purabaya Terminal
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD9980.5 Service industries--Quality control. H Social Sciences > HF Commerce > HF5415.335 Consumer satisfaction Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Noor Satriya
Date Deposited:	31 Jul 2026 06:17
Last Modified:	31 Jul 2026 06:17
URI:	http://repository.its.ac.id/id/eprint/140663

Actions (login required)

View Item