Feature Based Sentimental Analysis for Prediction of Mobile Reviews Using Hybrid Bag-Boost algorithm

Sentiment analysis or opinion mining is one of the major challenge of NLP (natural language processing). Business Analytics plays a key role in the current scenario with a perception that people wants to enhance their enterprise. In particular, these people rely on feedback of their goods to withstand the competition and knowledge mining that can give them an outstanding view into what to expect in the future. Few words or phrases may decide results or outcomes. As a majority of these people seek to boost their company in order to achieve full benefit by providing premium goods. In this aspect, sentiment analysis has gained a lot of interest in the current years. SA is an area of research of NLP that is used to classify a specific feature's opinion or perspective within a text. This paper is based on the different methods used to identify a particular text according to the opinions conveyed by the user's i.e. whether the overall sentiment of a individual is negative or positive or neutral. We are also looking at the two advance approaches adopted (feature classification pursued by polarity classification) along with the experimental results. Finally in this paper we compared 3 ML classification techniques 1) Logistic Regression, 2) Hybid Bag-Boost algorithm 3) SVM in which hybrid algorithm provides more accuracy compared to the other 3 ML algorithms. The Main objective of the proposed method is to predict the user reviews for choosing a best mobile using several classification Algorithms.