Fig 1 - uploaded by Adriana-Simona Mihaita
Content may be subject to copyright.
Source publication
Predicting how many travellers will choose a specific transport mode for daily commuting is always a challenging problem due to separate and large data sets, lack of integration and a significant over reliance mostly on surveying approaches. This paper presents a new approach for multi-modal transport choice prediction, via a hybrid structure of re...
Contexts in source publication
Context 1
... current methodology is applied over the New South Wales state in Australia, as represented in Fig. 1 which contains 198 LGA (local government areas) extending over a surface of 801, 150km 2 with a population of 8.166 million people as of September 2020, majority of which is concentrated in the city of Sydney (5.98 million people as of 2020 ...
Context 2
... different numbers of prediction targets Figure 10 shows the detailed RMSE vs. MAPE in predicting all combinations not including "Train" or "Bus", and the points are colored by the number of prediction targets. In this figure, each point presents the outcome of a certain regressor (such as RF, LR, MO-LSVR, and so on) when predicting a combination of predicted targets/modes (such as [Car as a Driver, Walk only] and [Work at home, Walk only, Car as passenger]). ...
Context 3
... both RMSE and MAPE are very small which is the best performance because the prediction tasks are the simplest. When predicting 2,3,4 modes at a time, there are two split bands of points and there is a trend that RMSE will decrease by increasing the prediction length while the MAPE will increase by increasing the number of prediction targets. Fig. 10 Detailed RMSE and MAPE in predicting combinations not including "Train" or "Bus", colored by the prediction feature ...
