Sentiment analysis on product reviews on twitter using Machine Learning Approaches

Social media have been obtained a wide growth within the recent years and have been interfering and affecting human lifestyle as well. This recent development has been resulted a huge collection of product reviews for a huge collection of consumers all round the world. These kinds of data collections are useful for future market predictions, to obtain customers' feedback, to provide product recommendations etc. Most of the above mentioned data collections are unstructured and handling those data bundles is a complex task. The act of examining and classifying the emotion or sentiment represented in any given review or text piece in order to establish whether the reviewer wishes to transmit positive or negative feelings is known as sentiment analysis. Sentiment analysis is one of the major part in natural language processing. In this study, twitter posts based on product reviews have been analyzed by using the support vector machine (SVM), logical regression and k-nearest neighbor machine learning algorithms and count vectorizer and tfidf mechanisms for converting texts into vectors in order to input data to the machine learning model. In this study, there are six main comparisons are done in order to find the most suitable "model vectorising" combinations from above mentioned. The highest accuracy score has been achieved by "Logistic Regression with Count Vectorizer" with an accuracy rate of 88.26%. Finally the true positive rates and true negative rates are going to calculate.