STB Sentiment Analysis Classification Multiclass Modeling Using Calibrated Classifier With SGDC Tuning As Basis and Sigmoid Method

Abstract

A set-top box (STB) is a device that converts digital signals into images and sounds that we can see on a regular analog television (TV). More and more people are looking for STB these days because the ordinary use of analog television (TV) will end in December 2022. The discontinuation of the normal use of analog television (TV) has generated a lot of positive, neutral, and negative sentiments. This sentiment data was obtained from social media Twitter using crawling data techniques. The feature extraction process in this study uses the TF-IDF method. In this study Stochastic Gradient Descent Classifier (SGDC) is used as a basis for determining the optimal method and comparing the SGDC tuning method with the Calibrated Classifier method. The test results show that the best optimization for this model is the Calibrated Classifier method with SGDC as its basis with an accuracy value of 80% on the test data and 100% on the training data. It shows that the Calibrated Classifier method can slightly improve the performance of testing and training on the SGDC Classifier has an accuracy value of 78.4% on the test data and 100% on the training data.