An Efficient Approach for Sentiment Analysis using Data Mining Algorithms
Milanjit Kaur, Kamini Joshi, Harmanjeet Singh
Abstract
Last few years have delivered a huge boom in the field of research in Sentiment Analysis, totally on incredibly subjective textual content kinds like films or products evaluations. This research work design a feature selection technique with Bagged Random Forest classification to predict the sentiments of news articles. This work represents comparison among C4. 5, random forest algorithm, bagged random forest algorithm and proposed algorithm by taking various parameters into consideration. Results are evaluated on the basis of various attributes namely correctly classified instances, incorrectly classified instances, error comparison. The basis of this comparison was done by taking various parameters into light such as mean absolute error, root mean squared error, relative absolute error, root relative squared error, average true positive rate, average false positive rate, recall and F-measure. These new results have proved that the proposed technique is having better results than the previous one.