Data generation has surged to an unprecedented level due to the rapid growth of information technology; in the last two years, data production has surpassed all previous recorded history in volume. Because of this, automated systems are now required to handle, understand, and use these data without the need for human involvement. Because of the widespread use of social media sites like Facebook, WhatsApp, and Twitter, sentiment analysis has become an important field of study. We analyzed 25 research publications in this study that highlight the shortcomings of current methods by focusing on machine learning techniques for sentiment analysis in mixed transliterated languages. Our assessment takes into account a number of variables, including journal sources, publication dates, performance measures, and the approaches’ numerical accomplishments. We also provide a thorough analysis of different approaches, highlighting their advantages and disadvantages. This paper concludes by discussing problems and future research topics for increasing the accuracy of sentiment analysis. Through a comparative analysis of the approaches, the motivation for the research is covered in detail.