The Impact of Principal Component Analysis Dimensionality Reduction on Sentiment Classification Performance Using Support Vector Machine

This study investigates the application of Principal Component Analysis (PCA) to enhance sentiment classification performance using the Support Vector Machine (SVM) algorithm. User reviews of the ChatGPT application from the Play Store were collected, preprocessed, and analyzed to identify the sentiment within the text (positive, negative, or neutral). The research follows the Knowledge Discovery in Databases (KDD) framework, starting with data selection, preprocessing, transformation, and applying PCA for dimensionality reduction. PCA was used to reduce the complexity of the high-dimensional text data, improving SVM's efficiency in sentiment classification. Evaluation results show that applying PCA led to an improvement in model performance, with accuracy increasing from 72.65% to 73.20%, precision from 71.58% to 72.24%, recall from 71.77% to 72.66%, and F1-score from 71.56% to 72.32%. Although the improvements were modest, the findings demonstrate that PCA effectively simplifies complex datasets and enhances SVM performance in sentiment classification, offering benefits in processing high-dimensional text data.