<![CDATA[Sentiment analysis involves using classification algorithms to analyze public opinions and feelings in text. Within the automobile industry, electric vehicles (EVs) stem from the circular economy and represent a novel technology under investigation in sentiment classification studies. The Support Vector Machine (SVM) algorithm is commonly used in this research due to its superior accuracy compared to other algorithms. The goal of this study is to apply SVM variable selection techniques to enhance sentiment analysis quality. Python is the programming language used to build the sentiment classification model, which involves feature selection using TF-IDF, training with cross-validation and grid search, evaluation using a confusion matrix, and storing the dataset in a MySQL database. The research focuses on the sentiment classification of 3000 public tweets about electric vehicles on Twitter. Through various scenarios, it was observed that the accuracy of sentiment classification varied depending on factors such as randomizing data, handling negation, and using different types of features like unigrams or bigrams. The highest accuracy achieved was 84% using a scenario with random data, negation handling, and unigram features. Overall, this research highlights the impact of randomizing data and selecting appropriate features on sentiment classification accuracy for electric vehicles on Twitter.]]>
Copyrights © 2023
