A person’s sentiment is rigorously influenced by his emotional feelings which is evoked from every single incident, occurring every day in his surroundings. In this case, the decision that he makes is greatly affected by his sentiment rather than facts. Sentimental behavior can be applied to many applications in health, business, education, etc. This paper proposes and develops a sentimental behavior based on machine learning algorithms for pre-processing feature selection, and classification that helps identify, extract, quantify the feelings to twitter social dataset. Based on the Sentimental behavior, an analytical prediction has been developed that can be used to understand the behavior of the customers or users. Conducting the testing process for three state-of-the-art algorithms, the unique methodology is devised based on these proposed algorithms (support vector machine (SVM), logistic regression (LR), and XGboost), our extensive experiments show that our approach has high accuracy. Based on the testing results, the accurate sentimental behavior detection algorithm is identified and recommended to be used for textual data in the future.