Table 5.

The average precision, ranking loss and coverage of IMMMLGP, Hum-mPloc, mGOF-loc, MKSVM, FSVM-KNR and out method when they are tested on datasets D3106 and D4802

	D3106			D4802
	RL	Cov	AP	RL	Cov	AP
IMMMLGP	0.4190	4.3030	0.5810	0.2436	4.9772	0.5725
Hum-mPloc	0.4906	5.3170	0.5790	0.3145	5.6830	0.5644
mGOF-loc	–	–	–	0.0606	3.0227	0.6482
MKSVM	0.1085	1.7193	0.7065	0.0662	2.9753	0.6889
FSVM-KNR	0.1071	1.7025	0.7108	0.0971	2.6339	0.6916
Our Method	0.0758	1.2848	0.7901	0.0637	3.0528	0.7414

Our method has great improvements on subcellular localization prediction than five currently available methods when tested on dataset D3106. The average precision of our method is 0.7901 which is 0.08 greater than the average precision of FSVM-KNR. The ranking loss and coverage of our method are lower than the values of other five methods. On dataset D4802, our method did not get the lowest ranking loss and coverage. However, the average precision of our method on dataset D4802 is the highest among those six methods with 0.7414. The best values are listed out with bold text