Search

Air quality plays a vital role in people's health, and air quality forecasting can assist in decision making for government planning and sustainable development. In contrast, it is challenging to multi-step forecast accurately due to its complex and nonlinear caused by both temporal and spatial dimensions. Deep models, with their ability to model strong nonlinearities, have become the primary methods for air quality forecasting. However, because of the lack of mechanism-based analysis, uninterpretability forecasting makes decisions risky, especially when the government makes decisions. This paper proposes an interpretable variational Bayesian deep learning model with information self-screening for PM2.5 forecasting. Firstly, based on factors related to PM2.5 concentration, e.g., temperature, humidity, wind speed, spatial distribution, etc., an interpretable multivariate data screening structure for PM2.5 forecasting was established to catch as much helpful information as possible. Secondly, the self-screening layer was implanted in the deep learning network to optimize the selection of input variables. Further, following implantation of the screening layer, a variational Bayesian gated recurrent unit (GRU) network was constructed to overcome the complex distribution of PM2.5 and achieve accurate multi-step forecasting. The high accuracy of the proposed method is verified by PM2.5 data in Beijing, China, which provides an effective way, with multiple factors for PM2.5 forecasting determined using deep learning technology. [ABSTRACT FROM AUTHOR]

• Novel multi-modal transport recommendation via graph neural networks. • Models travel behaviors with spatial and temporal data integration. • Introduces a contrastive learning framework to reduce bias. • CDHGNN: New de-biased recommendation framework tested in Beijing. • Demonstrates superiority over existing methods using 300,000 user records. Conventional uni-modal transportation recommendation systems focused on single modes of transportation are limited in providing satisfactory solutions since passengers often undertake journeys involving multiple modes. Multi-modal transportation recommendation systems are becoming increasingly popular within navigation applications. However, these systems face challenges from biased raw data, data sparsity and long-tail distribution, as well as complexities in representing large-scale graph structures, which collectively hinder their optimal performance. This study introduces a novel approach for enhancing multi-modal transportation recommendation systems: the C ontrastive D e-biased H eterogeneous G raph N eural N etwork (CDHGNN). By integrating contrastive learning, the model generates augmented samples to mitigate bias and overcome the data-skewing problem. The heterogeneous graph neural network adaptively captures temporal and spatial patterns among users and locations, as well as spatial adjacency and attribute relations, leading to enhanced representations of nodes, and consequently, improved model performance. The proposed method was evaluated using real-world data from over 300,000 users' records in Beijing over two months in 2018. The extensive experiments demonstrate that the approach outperforms four contemporary state-of-the-art methods. The results underscore the potential of the CDHGNN in large-scale city-level problems in practical applications, revealing a promising advancement for multi-modal transportation recommendation systems. [ABSTRACT FROM AUTHOR]

• This paper contributes to the understanding of the taste heterogeneity of travelers to pre-peak discount pricing strategy. • Naturalistic data from smart card data users in Beijing before and after a policy intervention is used in this regard. • Four groups of travelers with heterogeneous characteristics are classified by constructing a latent class model. • Different targeted policy recommendations for each group of travelers are proposed in order to improve the policy effect. Studies of travelers' response behavior to transportation demand management is receiving substantial attention among researchers and transport operators in recent years. While previous studies in this area have generally assumed that the sensitivity of travelers to different factors is homogeneous and relies on survey responses, which may be prone to self-reporting errors and/or subject to behavioral incongruence. Relying on naturalistic data, this paper aims to investigate the behavioral response to pre-peak discount pricing strategy in the context of the Beijing subway with a special focus on the heterogeneity among the travelers. Anonymous smart card data from 5946 travelers before and after the introduction of a peak avoidance policy in Beijing are used to construct a latent class choice model to capture the sensitivity to different factors and the associated taste heterogeneity of travelers. Given the passive nature of the data, the model can offer more realistic outputs. The results indicate that there is substantial heterogeneity in travelers' responses to the peak avoidance policy, and that they can be probabilistically allocated to four latent classes. For all classes of travelers, the decision to shift their departure to off-peak is affected by the monetary saving, the required change in departure time and the frequency of travel, but in different magnitudes. In particular, only two classes of travelers (who exhibit lower standard-deviation in pre-intervention departure time) show significant sensitivity to price changes indicating that the discount policies are more likely to be effective for these groups. The rest of travelers are largely price insensitive – warranting the need for non-monetary incentives as opposed to fare discounts. To the best of our knowledge, this study is the first to innovatively apply the LCC framework to analyze travelers' heterogeneous behavior using large-scale smart card data without socio-demographic information. The findings can provide guidance to the subway authority in devising differential peak avoidance policies targeted for different groups of users, which are likely to be more effective than the current 'one size fits all' approach. [ABSTRACT FROM AUTHOR]

User profiles, considered as one of the fundamental inputs of recommendation systems and customized services, can be rationally applied in the public transport domain to represent passengers' characteristics and behavioral preferences. A user profile of a bus passenger, termed as a bus passenger profile (BPP), is an assortment of labels containing passengers' travel features. This paper proposes a data-driven framework for developing BPP so as to provide guidance on how to create and estimate user profiles for bus passengers based on smart card data. The proposed method comprises three steps. (i) Data preprocessing aimed at extracting key information and preparing passenger profiling. (ii) A tag system aimed at storing the estimated travel features of passengers. (iii) Knowledge graphs aimed at connecting various BPP with semantic edges for practical application of prior knowledge in downstream tasks. The developed framework is implemented in a case study of the Beijing bus system. Deployment of the developed framework has demonstrated that it can satisfactorily develop BPP, while prior knowledge from the BPP-based knowledge graphs can benefit downstream tasks. [ABSTRACT FROM AUTHOR]

Lidar is a powerful active remote sensing technique to monitor the planetary boundary layer (PBL), and there have already been many algorithms to retrieve the planetary boundary layer height (PBLH) using aerosol lidars. However, the algorithm suitable for all atmospheric conditions doesn't exist. This study evaluates the capability of nine lidar algorithms under different pollution scenarios in terms of the instantaneous performance and the diurnal evolution, which are derived from classical theories and expressed as GM, LGM, CRGM, IPM, VAR, VGM, WCT, WCTD, and POLARIS in this paper. Based on the nine algorithms, the continuous PBLHs are obtained from the observation dataset of a Beijing urban site in 2017. In order to verify the lidar results, the statical comparison analysis with radiosonde is also presented. On the premise of the inversion hypothesis, such as sufficient mixing in the PBL, the consistency of lidar results was the best, with the difference being not more than 200 m. And what's different from our cognition is that the concentration of pollutants near the ground under the light pollution condition makes the lidar results tend to be consistent. In fact, LGM prefers to get higher PBLHs because of filtering the information near the ground, whereas POLARIS is apt to underestimate unless the actual morphology can be captured by the depolarization ratio. VAR and IPM are more vulnerable to the conspicuous stratification. GM, CRGM, and WCTD have better consistency with radiosonde and capture the evolution characteristic no matter in clean or polluted days. Compared with the clean days, PBLH from lidar algorithms under the polluted condition reduced about 500 m overall, as well as the standard deviation. Generally speaking, the comprehensive comparison provides a reference to choose the proper lidar algorithm when retrieving the PBLH. GM is appropriate for the detection of clean days, while CRGM is robust for polluted days. WCTD is more suitable for the judgment of PBLH in the morning and evening and will underestimate at noon. As to the situation with cloud, WCTD is also the best choice. • The nine algorithms are compared both internally and with radiosonde. • The difference of lidar results with sufficient mixing is not more than 200 m. • GM, WCT and WCTD are robust to meet the retrieval requirements. • WCTD is much better for the inversion of residual layer height at night. [ABSTRACT FROM AUTHOR]