PERFORMANCE ANALYSIS OF LOGISTIC REGRESSION ALGORITHM IN OPINION SEGMENTATION OF INDOSAT NETWORK SERVICE REVIEWS

In the era of the industrial revolution 4.0, where the use of network services has become a basic need and cannot be separated from daily activities, the massive number of network service users can be proven by the increasing number of people using digital platforms to search for information, express opinions or even just to communicate with each other, currently network services are available in the form of digital platforms that can be used to purchase network data packages or just to monitor the quality of network services, therefore this study aims to analyze user sentiment towards network services that have been launched by the Indosat provider based on the results of user reviews sourced from the digital platform using a machine learning approach and a logistic regression algorithm model to determine the segmentation of opinions that are widely expressed on the digital platform. The results of this study indicate that the logistic regression algorithm is able to analyze patterns of consumer characteristics with good accuracy in the algorithm model, and the results of the accuracy of the algorithm model in finding segmentation patterns in sentiment opinions reach an accuracy value of 85%, precision 81%, recall 77% and f1-score 79% to predict an opinion that has negative and positive sentiment during testing, then network speed, connection disruption and network data package prices are one of the factors that can influence an opinion regarding negative and positive sentiment.