Sentiment Analysis on Ajaib App Using the SVM Method

The rapid growth of investment applications has transformed trading accessibility, yet user dissatisfaction persists, particularly regarding transaction delays, technical issues, and inadequate customer support. This study addresses a research gap in sentiment analysis, specifically in the context of the Ajaib investment application, by employing a Support Vector Machine (SVM) model combined with lexicon-based labelling. The objective is to classify user-generated Google Play reviews into positive, negative, and neutral sentiments, providing actionable insights for service improvement. The research follows a structured methodology comprising data crawling, text pre-processing (cleaning, case folding, tokenization, stopword removal, and stemming), TF-IDF feature extraction, and supervised classification with SVM. Model validation utilises a 3×3 confusion matrix to calculate accuracy, precision, and recall, thereby ensuring a robust performance evaluation. Experimental results demonstrate that the SVM classifier achieves high accuracy in sentiment polarity classification, highlighting its suitability for text-based sentiment analysis in the financial domain. The distinct contribution of this research is its implementation of SVM for sentiment classification for Ajaib, offering a replicable framework for leveraging user feedback to enhance digital investment platforms. These findings contribute to the development of automated sentiment analysis systems that support data-driven decision-making for improving customer satisfaction.