Research Article

Data-Driven Simulation of Pedestrian Movement with Artificial Neural Network

Table 4

MTE and MDE of simulation cases with varied spatial matrix weights for the unidirectional flow scenario.

−0.1−0.2−0.3−0.4−0.5−0.6−0.7−0.8−0.9

MTE−0.10.12470.12390.11980.12090.11450.12100.11440.12500.1230
−0.20.12700.12630.11940.12380.12860.12020.11990.12720.1269
−0.30.12170.12910.12340.12720.12230.12210.12560.12050.1224
−0.40.12690.12320.12370.12280.12130.12340.11960.12370.1244

MDE−0.10.19560.18900.18180.18670.17380.18420.17140.19460.1879
−0.20.19580.19280.18710.19380.20220.18480.18710.19920.1980
−0.30.18630.20130.19130.19740.18980.19320.19510.18440.1868
−0.40.20020.19190.19140.19280.18670.19330.18390.19180.1925

For the bidirectional flow scenario, weight increment between the adjacent two bands is and weight increment from the central sectors to the adjacent outer ones is .