EXPLORING AIRBNB SATISFACTION IN INDONESIA’S TOP 5 TOURIST DESTINATION CITIES: A MACHINE LEARNING PERSPECTIVE

• Main Author: Abimanyu, Avisenna
• Format: Theses
• Language: Indonesia
• Subjects: Manajemen umum
• Online Access: https://digilib.itb.ac.id/gdl/view/75863
• Institution: Institut Teknologi Bandung

This research presents Airbnb customer satisfaction insights in Indonesia using machine learning models on Airbnb data. To develop the models, this research utilizes Airbnb data from Airbnb website as main data for machine learning process through web scraping and supporting analysis data from Central Bureau of Statistics Indonesia (BPS Indonesia). Customer reviews, listings scoring attributes (communication, cleanliness, accuracy, economic value, location, check in, total reviews, area, and price.), and cities are set as predictors. Author use different machine learning methods such as linear regression, sentiment analysis, LDA, and BERT each with its own respective predictor variables. For regression method author utilized 5 distinct data composition by city and for NLP based machine learning methods author utilized 8 distinct data composition: 3 parts by time period from overall data and 5 parts by city from overall data. To evaluate the model, author use root mean square error (RMSE) and mean absolute error (MAE) for linear regression. For LDA, author use log likelihood and perplexity score. For BERT, author use silhouette score. In the linear regression, by using weight to correlation before putting the important features into the model, we get different variations of the importance of listings scoring attributes for each city. Each city has good RMSE and MAE scores for predicting the result. From sentiment analysis, Airbnb users in Indonesia are satisfied with listings in Indonesia. With a comparison between satisfied and dissatisfied at 59:1. With Lombok, Surabaya, and 2011-2015 period have the highest difference between positive and negative reviews. From Topic Modelling, things that can improve negative customer satisfactions are improvement in places, facilities, cleanliness, and accommodation. Each city and each time period doesn’t have any big difference in the topics discussed, topics discussed are more or less about place, cleanliness, facilities, location, foods, and service. Place is the most common keyword found.