a new ml approach to sentiment analysis more standard form, for sentiment classification which falls under text classification is using many classification methods like naive bayes classifier, decision tree classifier, random forest classifier etc. Even though tere are many algorithms in scikit learn package but sometimes to understand the basic functionality of how the classification goes on, we generally opt for the popular way of statistical method, i.e averaging the effect of each word on the result and how to determine a strong word or a soft word. This algorithm gives exactly the meaning above and helps us to understand a new way of classification without actually using scientific packages and believing in simple "mean" method in obtaining consistent results after converting the dataset into word vectors.