A Joint Passenger Flow Inference and Path Recommender System for Deploying New Routes and Stations of Mass Transit Transportation

Abstract

In this work, a novel decision assistant system for urban transportation, called Route Scheme Assistant (RSA), is proposed to address two crucial issues that few former researches have focused on: route-based passenger flow (PF) inference and multivariant high-PF route recommendation. First, RSA can estimate the PF of arbitrary user-designated routes effectively by utilizing Deep Neural Network (DNN) for regression based on geographical information and spatial-temporal urban informatics. Second, our proposed Bidirectional Prioritized Spanning Tree (BDPST) intelligently combines the parallel computing concept and Gaussian mixture model (GMM) for route recommendation under users’ constraints running in a timely manner. We did experiments on bus-ticket data of Tainan and Chicago and the experimental results show that the PF inference model outperforms baseline and comparative methods from 41% to 57%. Moreover, the proposed BDPST algorithm's performance is not far away from the optimal PF and outperforms other comparative methods from 39% to 71% in large-scale route recommendations.

Traveling Transporter Problem: Arranging a New Circular Route in a Public Transportation System Based on Heterogeneous Non-Monotonic Urban Data

“…The workflow is depicted in Figure 1, including the preprocessing and computational phase. First of all, similar data preprocessing procedures as in other works [51,53] are adopted for grid-based storage as well as the construction of a grid-like graph for relevant urban features. However, the proposed method in the computational phase for solving TTP is made up of station recommendation and route planning.…”

“…Traditionally, methods that focus on designing new routes rely on origin-destination matrices to evaluate their expected efficacy [32,72]. In addition, various approaches such as transfer learning or other machine learning methods also have been introduced to deal with similar tasks [40,71,74]; among them, a state-of-the-art deep-neural-network-based PF inference module [51,53] that relies on data engineering with a route-affecting region (RAR) is proposed. The work [51,53] proposes a PF estimation model that emphasizes integrating the feature information of the route's influential region and combining the city's heterogeneous data, including but not limited to the point of interests, human mobility, and competitive and transferable relationship with existing routes.…”

“…In addition, various approaches such as transfer learning or other machine learning methods also have been introduced to deal with similar tasks [40,71,74]; among them, a state-of-the-art deep-neural-network-based PF inference module [51,53] that relies on data engineering with a route-affecting region (RAR) is proposed. The work [51,53] proposes a PF estimation model that emphasizes integrating the feature information of the route's influential region and combining the city's heterogeneous data, including but not limited to the point of interests, human mobility, and competitive and transferable relationship with existing routes. Thereafter, this neural-network-based inference module, with a grid-like graph that stores urban features and road network connectivity, is modified and adopted for evaluating the expected performance (PF) of arbitrary circular routes and constructing the graph needed in subsequent route planning.…”

“…In other words, one requires to re-organize and re-calculate so as to retain the exact effect on PF. In this work, the method to quantify the spatial relationship among routes and to evaluate the PF is based on a route-based inference module [51,53]. Admittedly, adopting an inference model will cause some error in the final PF; however, this adaptation and formulation do meet our targeted scenario in the real world since the current procedure relying on simulation also faces the uncertainty in PF.…”

“…In particular, as illustrated in Figure 1, the framework is composed of preprocessing and computational phases. First, based on a deep-neural-network-based PF inference module built on urban data [51,53], the PF of arbitrary routes can be accurately inferred. Next, in the second phase, stations are first recommended based on the Gaussian-Poisson Mixture Model with negative feedback, which is a modification that models the spatial attractiveness of PF for deployed stations and thereafter puts both PF of a single station and its herding effect into consideration.…”