earticle

영어

The analysis of social networking sites is a vast area of research as there are tremendous measures of records showing up in online networking. Predicting the comment patterns of users on these sites is a complex decision making process. This paper proposes a hybrid model of linear regression (PACE regression) and non linear regression (REP Tree) that predicts the likelihood of the comment volume, which a post may receive by analyzing the various features of the corresponding page, post and previous records of comment patterns of users. To mechanize the procedure, a model is built comprising of the crawler, data processor and information revelation module. The new hybridized model has improved the time and space complexity along with Accuracy by building a right sized tree using only significant features with low misclassification rate.