The internet has evolved into a forum for individuals to exchange their views, thoughts, opinions. Lot of social media applications like Facebook, Instagram have gained popularity where people can express their feelings, thoughts with audience. Twitter is one of social media application which is widely used by millions of peoples to express their views, opinions, thoughts to others. As per google reports in April 2024 X/Twitter have accounts over 500 million. As per records at least 500 million tweets per day which is about 6000 tweets per second or 35,000 tweets per minutes are being tweeted by users on twitter. These tweets include content such as entertainment news, sports news, trending topics, politics etc. Social media have made it easy for millions of peoples to communicate with one another online. Sentiment analysis is a technique that determine the emotional tone of digital text. It can be used to determine if the text message is positive, negative, neutral or irrelevant. In last two decades there have been plenty of research in this field of sentiment analysis. Processing of tweets make it challenging for researcher to research because tweet tweeted by people may contain some irrelevant words, emojis, hashtags, URLs, other signs. This paper offers an extensive examination and comparative analysis of existing technology for opinion mining by using various machine-learning algorithm such as SVM (support vector machine), Random Forest. We have discussed some applications, challenges, future scope, research gaps in twitter sentiment analysis.