Box Office Forecasting considering Competitive Environment and Word-of-Mouth in Social Networks: A Case Study of Korean Film Market

Table 8

Forecasting accuracy in terms of MAPE for each forecasting model and algorithm with different explanatory variables. , , and in the second row in each model denote that the MAPE of the corresponding model is lower than that of (S, MLR) at the significant level of 0.05, 0.1, and 0.2, respectively. Asterisks in the third, fourth, and fifth rows also indicate statistical significance in the aforementioned manner against (SC, MLR), (SW, MLR), and (SCW, MLR), respectively.

Variables	Screening (S)	Screening + competition (SC)	Screening + WOM (SW)	Screening + competition + WOM (SCW)
Algorithm	MLR	MLR	MLR	MLR	SVR	GPR	k-NN	ML average	Combination
Model	0.8383	0.4389	0.7482	0.3515	0.4420	0.3503	0.3175	0.3699	0.3121
		(47.64%)	(10.75%)	(58.07%)	(47.28%)	(58.21%)	(62.13%)	(55.87%)	(62.77%)
				(19.92%)	(−0.70%)	(20.19%)	(27.67%)	(15.72%)	(28.89%)
				(53.02%)	(40.93%)	(53.18%)	(57.57%)	(50.56%)	(58.29%)
					(−25.75%)	(0.33%)	(9.68%)	(−5.25%)	(11.20%)
									(15.63%)
Model	0.8391	0.6245	0.5325	0.4616	0.3343	0.2975	0.3731	0.3350	0.3274
		(25.58%)	(36.54%)	(44.99%)	(60.16%)	(64.54%)	(55.54%)	(60.08%)	(60.98%)
				(26.08%)	(46.46%)	(52.35%)	(40.25%)	(46.36%)	(47.57%)
				(13.30%)	(37.21%)	(44.12%)	(29.93%)	(37.09%)	(38.51%)
					(27.58%)	(35.55%)	(19.18%)	(27.44%)	(29.08%)
									(2.26%)
Model T	0.5501	0.4702	0.4583	0.3681	0.2666	0.2807	0.3123	0.2865	0.2599
		(14.54%)	(16.69%)	(33.10%)	(51.54%)	(48.98%)	(43.23%)	(47.92%)	(52.76%)
				(21.72%)	(43.30%)	(40.30%)	(33.58%)	(39.06%)	(44.72%)
				(19.70%)	(41.83%)	(38.75%)	(31.86%)	(37.48%)	(43.29%)
					(27.56%)	(23.73%)	(15.15%)	(22.15%)	(29.39%)
									(9.30%)