Predicting the Currency Market in Online Gaming via Lexicon-Based Analysis on Its Online Forum

Table 5

Experimental results of predicted WoW token prices fluctuation.

Learning method		Gradient boosting			Random forest			Support vector machine
Region	Learning days	Accuracy (%)	1-score	MCC	Accuracy (%)	1-score	MCC	Accuracy (%)	1-score	MCC
North America region	3 days	75.65%	0.7576	0.5731	69.13%	0.6984	0.4799	64.13%	0.6432	0.3445
	5 days	79.56%	0.7955	0.5921	76.08%	0.7784	0.5704	74.48%	0.7539	0.318
	7 days	82.55	0.8255	0.6372	78.26%	0.793	0.5768	76.95%	0.7765	0.3106
	12 days	80.65%	0.8045	0.6009	78.69%	0.7858	0.579	78.69%	0.7996	0.3773
Europe region	3 days	69.56%	0.6957	0.3917	71.52%	0.7138	0.4375	75.43%	0.7966	−0.0219
	5 days	71.73%	0.7174	0.4346	73.91%	0.7367	0.4907	70.65%	0.6225	0.2742
	7 days	77.17%	0.7718	0.5429	77.39%	0.77	0.5723	71.3%	0.6346	0.2989
	12 days	81.52	0.815	0.6322	77.6%	0.7791	0.5779	73.91%	0.6656	0.3416