Mass Rapid Transit System Passenger Traffic Forecast Using a Re-Sample Recurrent Neural Network

Table 2

The performance of models with different weights of resample rate (congestion, moderate, light, and normal) and different hidden sizes.

model	Test Accuracy	precision(severe, moderate,light,seat available)				recall(severe, moderate,light,seat available)				F1-score(severe, moderate,light,seat
Weighs(0.25,0.25,0.25,0.25)HiddenSize 200	0.792	0.88	0.69,	0.71,	0.92	0.88,	0.77,	0.70,	0.82	0.88,	0.73,	0.70,	0.87
Weights(0.2,0.2,0.2,0.4)Hiddensize 200	0.81	0.86,	0.72,	0.68,	0.90	0.81,	0.73,	0.69,	0.90	0.83,	0.72,	0.69,	0.90
Weights(0.15,0.15,0.2,0.5) Hiddensize 200	0.837	0.89,	0.70,	0.71,	0.92	0.80,	0.76,	0.69,	0.93	0.84,	0.73,	0.70,	0.92
Weights(0.02,0.023,0.097,0.086)Hiddensize 200	0.92	0.57,	0.56,	0.73,	0.95	0.40,	0.43,	0.63,	0.98	0.47,	0.49,	0.68,	0.97
Weights(0.2,0.2,0.2,0.4)Hiddensize 150	0.811	0.81,	0.70,	0.74,	0.90	0.83,	0.70,	0.68,	0.92	0.82,	0.70,	0.71,	0.92
Weights(0.15,0.15,0.2,0.5) Hiddensize 150	0.836	0.83,	0.68,	0.73,	0.94	0.79,	0.79,	0.74,	0.90	0.81,	0.73,	0.74,	0.92
Weighs(0.25,0.25,0.25,0.25)HiddenSize 150	0.827	0.94,	0.72,	0.78,	0.89	0.84,	0.86,	0.70,	0.90	0.89,	0.78,	0.74,	0.90
Weights(0.02,0.023,0.097,0.86)Hiddensize 150	0.937	0.69,	0.68,	0.78,	0.96	0.45,	0.57,	0.68,	0.99	0.55,	0.62,	0.73,	0.97
Weights(0.2,0.2,0.2,0.4)Hiddensize 250	0.832	0.88	0.67,	0.74,	0.95	0.84,	0.81,	0.68,	0.92	0.86,	0.73,	0.71,	0.93
Weights(0.15,0.15,0.2,0.5) Hiddensize 250	0.86	0.91,	0.74,	0.78,	0.91	0.81,	0.78,	0.74,	0.95	0.80,	0.72,	0.79,	0.95
Weighs(0.25,0.25,0.25,0.25)HiddenSize 250	0.791	0.82,	0.66,	0.79,	0.92	0.82,	0.76,	0.72,	0.88	0.82,	0.70,	0.75,	0.90
Weights(0.02,0.023,0.097,0.86)Hiddensize 250	0.93	0.79,	0.53,	0.71,	0.97	0.55,	0.43,	0.71,	0.98	0.65,	0.48,	0.71,	0.97
Without resample Hidden size 250	0.927	0.56,	0.78,	0.68,	0.96	0.62,	0.54,	0.67,	0.97	0.59,	0.64,	0.67,	0.97