Social media archives have enormous amounts of a wide range of unstructured data kinds. Among them are text data, audio, video, and visual media. It also includes sentiments, medical information, debate topics, and client testimonials. The data also includes client evaluations of goods and services. There are countless amounts of online reviews. Because of this, it may be challenging for a potential merchant to analyze them. It also makes it difficult for the product's creator to keep track of and manage user reviews. Sentiment analysis is a technique that helps with the challenging process by looking at the emotions expressed in so many online evaluations. Above all, sentiment analysis yields beneficial outcomes based on facts, enabling you to decide for your organization from the most important feelings present in social media. Sentiment analysis (SA) is a method of natural language processing that seeks to identify emotions related to a given topic and extract views about that topic from a vast corpus of data. The objective of this work is to examine the sentiment analysis technique like bag-of-word, Word distribution – inverse document frequency, Vader sentiment Analysis and evaluate the performance of classification algorithms for the analysis of twitter poco customer review sentiments. The classification techniques Tree Logistic Model Tree (LMT), Lazy Bayesian Rules (LBR), Hoeffding tree classifier, and Naive Bayes classifier are employed in this work to examine consumer sentiments. This work determines which algorithms are most appropriate for the analysis of text data.

Twitter Sentiment Analysis, Bag-of-Words method, VADER sentiment Analysis, inverse document frequency method.