Performance Assessment of Multiple Classifiers Based on Ensemble Feature Selection Scheme for Sentiment Analysis

Table 5

Results of the proposed model for electronics review data set.

Method	Classifier
	SVM			MNB			KNN			ME
	Prec	Recall	F- Score	Prec	Recall	F-Score	Prec	Recall	F-Score	Prec	Recall	F-Score
Unigram	86.9	83.2	85.0	85.7	86.6	86.14	82.3	80.1	81.18	82.1	79.8	80.93
Bigram	81.5	80.1	80.79	82.5	80.2	81.33	84.5	81.2	82.81	78.2	76.2	77.18
Unigram+	87.2	84.4	85.77	85.2	87.4	86.28	86.2	82.7	80.41	83.1	80.4	81.72
Bigram
CompIG	88.3	87.1	87.69	89.2	86.0	87.96	87.7	84.5	85.07	85.7	81.9	83.75
CompCHI	86.7	82.5	84.54	87.9	85.4	86.63	83.9	81.2	82.52	86.2	83.6	84.88
CompGI	88.1	84.8	86.41	88.2	85.7	86.93	88.6	83.4	84.92	87.3	85.5	86.39
Average	86.6	83.5	85.01	86.3	85.4	86.33	85.53	82.18	83.81	83.76	82.10	84.75