Impact of GIS and Naive Bayes on Ride-Hailing User Satisfaction

Abstract

In the digital age, consumer behavior has shifted towards online media, facilitated by the affordability, speed, and advancement of super apps like GOJEK. These apps offer multiple services in a single platform, transforming how people fulfill their daily needs. GOJEK, with its extensive services, has emerged as a dominant player in Indonesia, commanding a large user base. Despite competition from Grab, GOJEK's dominance underscores its efficient integration of the super app concept. A research study aims to assess and analyze the satisfaction levels of GOJEK users in Indonesia using the Naive Bayes method. The study seeks to understand user interactions with super apps and contribute to the evolving digital landscape of Indonesian society. The integration of GIS and data analysis techniques presents a promising avenue for improving user satisfaction and the overall success of ride-hailing services. As the industry continues to expand and evolve, the insights derived from this research could be crucial in shaping the future of the ride-hailing industry. The research employs the Naive Bayes classification method to gauge the probability of satisfaction among GOJEK users in Indonesia. The analysis of satisfaction probability is conducted to understand the extent to which users are content with the service. The study utilizes a dataset comprising 20,000 GOJEK user records from Indonesia. The dataset preparation involves the removal of missing values, duplicate entries, and data transformation to incorporate a label column that classifies user satisfaction based on the scores provided. The computations of prior, likelihood, and posterior probabilities using the Naive Bayes method reveal that the probability of GOJEK user satisfaction in the "Satisfied" category surpasses that in the "Dissatisfied" category. The model evaluation demonstrates a high level of accuracy, with accuracy and precision values reaching 0.645, a Recall value of 1.0, and an F1-score of 0.784. These outcomes indicate that the Naive Bayes model effectively predicts GOJEK user satisfaction based on the scores given.