A Novel Feature Selection Technique for Text Classification Using Naïve Bayes

Table 9

(a) Comparison of proposed method with greedy search. (b) Comparison of the proposed method with CFS.

(a)

Datasets	Wrapper forward search greedy		NB + FS-CHICLUST
	Execution time	Classification accuracy	Execution time	Classification accuracy
CNAE-9	53.76 min	54%	0.81 min	68%
Hotel	44.89 min	60%	3.92 min	87%
Gender	3.10 min	59%	0.06 min	69%
Prosncons	71.82 min	51%	1.70 min	76%
CookWare	48.19 min	53%	3 min	80%
MyMail	120.41 min	85%	5.10 min	93%
Reuters	119.30 min	19%	3.80 min	76%
Computers	65.75 min	40%	4.25 min	93%
Flipkart	105.48 min	67%	4.12 min	84%
SpamHam	191.41 min	14%	5.15 min	95%
Books	42 min	33%	3.85 min	93%
DBTotal	50.52 min	60%	8.1 min	90%

(b)

Datasets	Filter (CFS)		NB + FS-CHICLUST
	Execution time	Classification accuracy	Execution time	Classification accuracy
CNAE-9	85.80 min	68%	0.81 min	68%
Gender	22.02 min	55%	0.06 min	69%
Prosncons	293.40 min	70%	1.70 min	76%
CookWare	308.34 min	73%	3 min	80%
DBTotal	24 min	75%	8.1 min	90%
Books	348 min	53%	3.85 min	93%
Hotel	382.4 min	60%	3.92 min	87%
Computers	392 min	67%	4.25 min	93%