ANALISIS SENTIMEN KEUANGAN (DATA FIQA AND FINANCIAL PHRASEBANK) MENGGUNAKAN ALGORITMA LOGISTIC REGRESSION DAN SUPPORT VECTOR MACHINE

Abstract

Finance is a very vital sector in a Company and institution because it has a very important strategic role in creating a conducive environment, especially for the improvement of the national economy. Through a combination of FiQA and Financial PhareBank text datasets, an analysis of positive, negative and neutral sentiments related to finance is carried out that can be taken into consideration to make a policy in the financial sector or context in achieving this strategic role. Application of sentiment analysis using hyperparameter tuning in Logistic Regression and Support Vector Machines algorithms, with TF-IDF and Smote weighting on training data. The best model results of 70.70% accuracy on the Support Vector Machine algorithm during model training using training data that is not done Smote class imbalance.