Forecasting ridership for a metropolitan transit authority

The recent volatility in gasoline prices and the economic downturn have made the management of public transportation systems particularly challenging. Accurate forecasts of ridership are necessary for the planning and operation of transit services. In this paper, monthly ridership of the Metropolitan Tulsa Transit Authority is analyzed to identify the relevant factors that influence transit use. Alternative forecasting models are also developed and evaluated based on these factors—using regression analysis (with autoregressive error correction), neural networks, and ARIMA models—to predict transit ridership. It is found that a simple combination of these forecasting methodologies yields greater forecast accuracy than the individual models separately. Finally, a scenario analysis is conducted to assess the impact of transit policies on long-term ridership.     

Error and optimism bias in toll road traffic forecasts

Traffic forecasts are employed in the toll road sector, inter alia, by private sector investors to gauge the bankability of candidate investment projects. Although much is written in the literature about the theory and practice of traffic forecasting, surprisingly little attention has been paid to the predictive accuracy of traffic forecasting models. This paper addresses that shortcoming by reporting the results from the largest study of toll road forecasting performance ever conducted. The author had access to commercial-in-confidence documentation released to project financiers and, over a 4-year period, compiled a database of predicted and actual traffic usage for over 100 international, privately financed toll road projects. The findings suggest that toll road traffic forecasts are characterised by large errors and considerable optimism bias. As a result, financial engineers need to ensure that transaction structuring remains flexible and retains liquidity such that material departures from traffic expectations can be accommodated.

Strategic planning support for road safety measures based on accident data mining

When actions and measures to increase road safety are to be planned by the police and local authorities, it is necessary to consider the specific accident circumstances as well as their historical, current, and predicted course. In particular, combinations of accident circumstances not contained in existing police statistics are often neglected, but may nevertheless be relevant, e. g., due to an increasing frequency. In order to identify these undiscovered interesting combinations, we propose a framework to support strategic planning of road safety measures based on several consecutive data mining stages. The scope, type, and location of road safety measures must be planned at a strategic level several months in advance to be fully effective. Therefore, it is essential to investigate and predict the accident circumstances and the temporal changes in their frequency comprehensively. Only with the knowledge, e. g., about the temporal pattern, locations, conditions of roads or speeds, meaningful actions can be derived. The embedded data mining approaches, i. e., frequent itemset mining, time series clustering, time series classification, forecasting, and scoring, are carefully selected, coordinated, and aligned. As a result, the framework provides police users with information about circumstances of accidents that are of interest in the future and presents their previous temporal and local patterns in a dashboard. In this study, the framework is applied in four different geographical regions. Thereby, default parameter settings for all approaches are found that are particularly suitable for the framework to investigate novel geographic regions.     

最优组合预测方法在内河港口吞吐量预测中的应用

港口吞吐量的预测是港口规划过程中最为基础也最为关键的一步,只有对港口吞吐量做出准确、稳定的预测,才能做出科学合理的港口发展规划。由于内河港吞吐量历史数据有限,文中采用GM(1,1)和Verhulst模型的最优组合模型对港口吞吐量进行预测。该组合模型充分利用GM(1,1)模型“少数据,短期预测准确”的优点,又针对GM(1,1)预测量的无限增大趋势,引入了Verhulst模型进行组合修正,进而提高预测值的准确、稳健性。     

港口吞吐量计量预测分析

通过应用国际航运的派生需求理论和计量经济分析方法,采用计量经济学中时间趋势、对数等变量分析手段,新设计了一个能精确预测港口吞吐量的经济计量方程。通过使用该计量方程,并引用厦门港吞吐量历年数据,采用两种不同的计量技术方法对厦门港2005～2008年的集装箱吞吐量进行预测,两种方法的预测结果差异很小。论文同时对两种预测方法获得的预测结果进行了误差分析,并计算出误差结果。     

Local online kernel ridge regression for forecasting of urban travel times

Accurate and reliable forecasting of traffic variables is one of the primary functions of Intelligent Transportation Systems. Reliable systems that are able to forecast traffic conditions accurately, multiple time steps into the future, are required for advanced traveller information systems. However, traffic forecasting is a difficult task because of the nonlinear and nonstationary properties of traffic series. Traditional linear models are incapable of modelling such properties, and typically perform poorly, particularly when conditions differ from the norm. Machine learning approaches such as artificial neural networks, nonparametric regression and kernel methods (KMs) have often been shown to outperform linear models in the literature. A bottleneck of the latter approach is that the information pertaining to all previous traffic states must be contained within the kernel, but the computational complexity of KMs usually scales cubically with the number of data points in the kernel. In this paper, a novel kernel-based machine learning (ML) algorithm is developed, namely the local online kernel ridge regression (LOKRR) model. Exploiting the observation that traffic data exhibits strong cyclic patterns characterised by rush hour traffic, LOKRR makes use of local kernels with varying parameters that are defined around each time point. This approach has 3 advantages over the standard single kernel approach: (1) It allows parameters to vary by time of day, capturing the time varying distribution of traffic data; (2) It allows smaller kernels to be defined that contain only the relevant traffic patterns, and; (3) It is online, allowing new traffic data to be incorporated as it arrives. The model is applied to the forecasting of travel times on London’s road network, and is found to outperform three benchmark models in forecasting up to 1 h ahead.     

Forecasting the short-term metro passenger flow with empirical mode decomposition and neural networks

Short-term passenger flow forecasting is a vital component of transportation systems. The forecasting results can be applied to support transportation system management such as operation planning, and station passenger crowd regulation planning. In this paper, a hybrid EMD-BPN forecasting approach which combines empirical mode decomposition (EMD) and back-propagation neural networks (BPN) is developed to predict the short-term passenger flow in metro systems. There are three stages in the EMD-BPN forecasting approach. The first stage (EMD Stage) decomposes the short-term passenger flow series data into a number of intrinsic mode function (IMF) components. The second stage (Component Identification Stage) identifies the meaningful IMFs as inputs for BPN. The third stage (BPN Stage) applies BPN to perform the passenger flow forecasting. The historical passenger flow data, the extracted EMD components and temporal factors (i.e., the day of the week, the time period of the day, and weekday or weekend) are taken as inputs in the third stage. The experimental results indicate that the proposed hybrid EMD-BPN approach performs well and stably in forecasting the short-term metro passenger flow.     

Traffic volume forecasting based on radial basis function neural network with the consideration of traffic flows at the adjacent intersections

The forecasting of short-term traffic flow is one of the key issues in the field of dynamic traffic control and management. Because of the uncertainty and nonlinearity, short-term traffic flow forecasting could be a challenging task. Artificial Neural Network (ANN) could be a good solution to this issue as it is possible to obtain a higher forecasting accuracy within relatively short time through this tool. Traditional methods for traffic flow forecasting generally based on a separated single point. However, it is found that traffic flows from adjacent intersections show a similar trend. It indicates that the vehicle accumulation and dissipation influence the traffic volumes of the adjacent intersections. This paper presents a novel method, which considers the travel flows of the adjacent intersections when forecasting the one of the middle. Computational experiments show that the proposed model is both effective and practical.     

无偏灰色预测在列车自动驾驶系统中的应用

列车自动驾驶系统是列车控制系统研究中的一部分,可以按照目标速度自动完成运输任务,并且最大程度地节约能源、提高旅行速度。无偏GM（1,1）模型克服传统灰色预测模型存在灰色偏差与抗干扰能力弱的局限性,提高速度预测值的精度。通过算例表明,该方法具有可靠的预测结果,能有效地指导列车自动驾驶。     

Traffic flow forecasting based on a hybrid model

Real-time traffic flow forecasting is of great importance in the development of advanced traffic management systems and advanced traveler information systems. Traffic flow is evaluated using time series, and the Autoregressive Integrated Moving Average (ARIMA) model has been commonly used for determining the regression-type relationship between historical and future data. However, the performance of the ARIMA model is limited by the difficulty of capturing nonlinear patterns and the challenges of diagnosing permanent white noises. Hence, a hybrid method of ARIMA-EGARCH-M-GED was developed with the intent to address those limitations. It combines the linear ARIMA model with a nonlinear model of Exponent Generalized Autoregressive Conditional Heteroscedasticity (EGARCH) to capture heteroscedasticity (the variance of random error varying across the data) of traffic flow series. EGARCH in Mean (EGARCH-M), which corrects the expression of conditional variance by connecting the conditional mean directly with the variance, was introduced to better restrain the influence of abnormal data. Moreover, the tail of the generalized error distribution (GED) is better than that of the normal distribution in characterizing the features of time series, especially heteroscedasticity of residual sequences. Data collected from an interstate highway (I-80 in California) with a sampling period of 5 minutes were used to evaluate the performance of the proposed model. The results from the hybrid model were compared with ARIMA, an artificial neural network, and a K-nearest neighbor model. The results showed that the hybrid model outperformed the other methods in terms of accuracy and reliability. Overall, the proposed model performed well in tracking the features of measured data and controlling the impact of abnormal data.