Abstract
From 1980, travel time of Metro Manila bus services have increased by about 64%. To address this increase in travel time and improve service reliability, different studies have proposed providing travel time information and analyzing components that affect total travel time. This paper analyzed the impact of passenger movements and counts to dwell time using a 3-27-1 Artificial Neural Network (ANN). Results show that the ANN model and the OLS linear regression model with similar variables both resulted to high errors in prediction. Further statistical analysis showed that there have been no significant differences between actual values and ANN-predicted values. This suggests that ANN is an acceptable model for dwell time. However, testing results showed that passenger counts only explain about 82% of the variability. As a suggestion, further investigation of other variables (aside from passenger counts) that may affect bus dwell time should be considered.
