Penggunaan Metode Naïve Bayes Classifier pada Analisis Sentimen Penilaian Masyarakat Terhadap Pelayanan Rumah Sakit di Malang

Abstrak

Peran rumah sakit dalam kehidupan masyarakat sangatlah penting terkait tingkat kepuasan Masyarakat terhadap pelayanan, fasilitas, dan aspek lainnya. Opini dan penilaian masyarakat turut menjadi penilaian terhadap kinerja pelayanan rumah sakit. Pada Google Maps Reviews banyak ulasan dari berbagai rumah sakit.Penilaian yang sangat besar dapat kita lihat pada Google Maps Reviews akan memakan waktu bagi masyarakat. Keluhan-keluhan Masyarakat disekitar penulis terhadap pelayanan rumah sakit di Malang menjadikan penilaian pelayanan rumah sakit di Malang menjadi objek dari penelitian dasar ini. Penulis memanfaatkan algoritma Naïve Bayes Classifier dan Cross Validation untuk mengkategorikan penilaian berdasarkan sentimen positif dan negatif serta  aspek agar mempermudah pengkategorian. Aspek yang dipergunakan tersebut adalah aspek penanganan, fasilitas, administrasi, dan biaya. Penulis juga menggunakan analisis Root Cause untuk mempermudah masyarakat dan pihak terkait dalam menemukan masalah dan rekomendasi pemecahan masalah. Awalnya data di proses dengan  text preprocessing lalu pembobotan kata TF-IDF, pelabelan data, penerapan algoritma Naïve Bayes Classifier dan mengambil sentimen negatif untuk menentukan Root Cause. Hasil pengujian dengan menggunakan Cross Validation dengan fold k-9 memiliki nilai accuracy 82,97% , precision sebesar 83,13%, recall 82,93%, dan f-measure sebesar 82,92%. Hasil uji dengan menggunakan 20% data tes diperoleh akurasi 90%.

Abstract

The role of hospitals in society is crucial in terms of the level of satisfaction that the community derives from their services, facilities, and other aspects. Public opinions and assessments also contribute to evaluating hospital service performance. On Google Maps Reviews, there are numerous reviews from various hospitals. A significant evaluation can be observed on Google Maps Reviews, which might take time for the community. The complaints of the community around the writer regarding the hospital services in Malang make the assessment of hospital services in Malang the subject of this basic research. The author utilizes the Naive Bayes Classifier algorithm and Cross Validation to categorize assessments based on positive and negative sentiments, as well as 4 aspects to facilitate categorization. The author also employs Root Cause analysis to aid the public and relevant parties in identifying issues and providing problem-solving recommendations. After processing the data through text preprocessing and TF-IDF word weighting, data labeling, applying the Naive Bayes Classifier algorithm, and extracting negative sentiments to determine the Root Cause in negative hospital sentiments. Based on this process, applying Cross Validation with k-9 folds yields the highest values: an accuracy of 82.97%, precision of 83.13%, recall of 82.93%, and an f-measure of 82.92%. Through the sentiment classification and Cross Validation process, the accuracy results in 90% for hospital reviews with the highest number of assessments divided into 2 sentiments and 4 aspects: positive and negative sentiments, as well as treatment, facilities, administration, and costs.