notesum.ai

Published at November 27

SPTTE: A Spatiotemporal Probabilistic Framework for Travel Time Estimation

cs.LG

Released Date: November 27, 2024

Authors: Chen Xu¹, Qiang Wang¹, Lijun Sun¹

Aff.: ¹Not specified

Arxiv: http://arxiv.org/abs/2411.18484v1

Refer to caption

Model	Chengdu					Harbin
Model	RMSE	MAE	MAPE(%)	CPRS		RMSE	MAE	MAPE(%)	CPRS
DeepTTE	$180.96$	$130.02$	$17.15$	—		$224.31$	$162.52$	$18.31$	—
HierETA	$155.17$	$111.29$	$14.69$	—		$187.95$	$136.41$	$15.60$	—
MulT-TTE	$149.92$	$105.22$	$13.69$	—		$178.33$	$129.77$	$14.83$	—
DeepGTT	$165.15$	$118.68$	$15.64$	$1.46$		$199.25$	$143.95$	$16.40$	$1.56$
GMDNet	$151.50$	$107.75$	$13.96$	$1.30$		$176.78$	$127.81$	$14.49$	$1.41$
ProbETA	${138.36}$	${97.45}$	${12.54}$	${1.19}$		${159.68}$	${115.79}$	${13.17}$	${1.29}$
SPTTE	134.79	94.81	12.25	1.17		157.60	113.62	12.96	1.26
Improvement	${10.09\%}$	${9.89\%}$	${10.53\%}$	${10.00\%}$		${10.85\%}$	${11.10\%}$	${10.57\%}$	${10.64\%}$