Research on Railway Passenger Volume Forecast Based on the Spline Interpolation and IPSO-Gradient Difference Acceleration Rule

Table 2

Forecast values of the railroad passenger traffic in Beijing for different forecasting methods.

Forecast year	BP neural network prediction method		Holt exponential smoothing method		Average weighted portfolio method		IPSO-redifferential acceleration law method
	Forecast value (million people)	Absolute value of prediction errors (%)	Forecast value (million people)	Absolute value of prediction errors (%)	Forecast value (million people)	Absolute value of prediction errors (%)	Forecast value (million people)	Absolute value of prediction errors (%)
2017	14195	2.323	13996	0.887	14096	1.605	13777	0.694
2018	14922	4.548	14613	2.382	14768	3.465	14272	0.007
2019	15495	5.017	15229	3.212	15362	4.115	14936	1.227