In the current digital environment, almost all business has gone online. The customer’s purchase is influenced by the product’s appeal and the reviews by other customers. Whether favorable or unfavorable, the remarks will undoubtedly help business people in their decision-making regarding marketing. This increases the need for sentiment analysis algorithms. There are numerous machine learning and deep learning methods available for this type of investigation. Here, the features taken into account for analysis have a major impact on the accuracy of the outcome. The suggested study makes use of a mean-based polarity embedded feature selection strategy for SVM, increasing sentiment analysis’s precision. The outcome shows that the accuracy of SVM for sentiment analysis will increase when polarity is used for feature selection.