Home-based trip end models — A comparison between category analysis and regression analysis procedures

The trip end models which have been used in past transportation studies are briefly summarised. Problems associated with the use of zone-based models are outlined and reasons are given to support the development of models at the household rather than zonal level.It is suggested that recent developments which have taken place in household-based models have not been entirely logical. In particular, arguments between regression models and category analysis models have been confused with the use of aggregate (zonal) as against disaggregate (household) data — regression models being associated with the use of zonal data and category analysis models with household data. Misunderstood arguments and false notions regarding sample sizes have directed attention from the regression analysis approach.A detailed comparison of the category analysis and regression analysis methods for developing household-based trip end models is given. Both methods have been applied using data from the Monmouthshire Land Use Transportation Study. The regression results reported are from a very preliminary analysis and contain a number of anomalies, although it is thought that sufficient work has been done to provide an objective evaluation of the two methods.It is recommended that the household regression approach should be further investigated since it has advantages as a modelbuilding procedure and makes better use of sample data. A certain amount of categorisation of household types is necessary and the investigations would attempt to determine the best balance between categorisation and regression fitting. Further development will be restricted if the trend towards minimum sample sizes of about 1000 households is continued. Larger samples should be taken in certain circumstances to pursue development work.     

Ridership estimation for short-range transit planning

The objective of this research was to develop a simple transit ridership estimation model system for short-range planning. The main feature of the model system is that it exploits knowledge of transit link volumes which are obtained readily from on-off counts. Extensive use is made of default values for model parameters, taken directly from the transportation literature. The remaining parameters can be derived easily from generally available land-use and socioeconomic data. Expensive household surveys and time-consuming model calibrations are not required. A sequence of simple trip generation, trip distribution and modal split models generate trip-purpose specific transit trip tables, denoted as “trial” trip tables. These trip tables and observed transit link volumes are used in a linear programming model which serves as a correction mechanism. The gain in accuracy is achieved by using the ridership information contained in the transit link volumes. The corrected trip tables may be used in a pivot-point analysis to estimate changes in ridership and revenue. The results of a test application of the model system indicate that it can generate accurate ridership estimates when reliable transit link volumes are available from on-off counts, and when the trial transit trip tables as derived from the first three component models are reasonably accurate.     

Path flow estimator for planning applications in small communities

This paper presents an alternative planning framework to model and forecast network traffic for planning applications in small communities, where limited resources debilitate the development and applications of the conventional four-step travel demand forecasting model. The core idea is to use the Path Flow Estimator (PFE) to estimate current and forecast future traffic demand while taking into account of various field and planning data as modeling constraints. Specifically, two versions of PFE are developed: a base year PFE for estimating the current network traffic conditions using field data and planning data, if available, and a future year PFE for predicting future network traffic conditions using forecast planning data and the estimated base year origin–destination trip table as constraints. In the absence of travel survey data, the proposed method uses similar data (traffic counts and land use data) as a four-step model for model development and calibration. Since the Institute of Transportation Engineers (ITE) trip generation rates and Highway Capacity Manual (HCM) are both utilized in the modeling process, the analysis scope and results are consistent with those of common traffic impact studies and other short-range, localized transportation improvement programs. Solution algorithms are also developed to solve the two PFE models and integrated into a GIS-based software called Visual PFE. For proof of concept, two case studies in northern California are performed to demonstrate how the tool can be used in practice. The first case study is a small community of St. Helena, where the city’s planning department has neither an existing travel demand model nor the budget for developing a full four-step model. The second case study is in the city of Eureka, where there is a four-step model developed for the Humboldt County that can be used for comparison. The results show that the proposed approach is applicable for small communities with limited resources.     

Comparison of pedestrian trip generation models

Using Poisson regression and negative binomial regression, this paper presents an empirical comparison of four different regression models for the estimation of pedestrian demand at the regional level and finds the most appropriate model with reference to the National Household Travel Survey (NHTS) 2001 data for the Baltimore (USA) region. The results show that Poisson regression seems to be more appropriate for pedestrian trip generation modeling in terms of χ² ratio test, Pseudo R², and Akaike's information criterion (AIC). However, R² based on deviance residuals and estimated log‐likelihood value at convergence confirmed the empirical studies that negative binomial regression is more appropriate for the over‐dispersed dependent variable than Poisson regression. Copyright © 2011 John Wiley & Sons, Ltd.     

Evaluating light rail sketch planning: actual versus predicted station boardings in Phoenix

In recent years, transit planners are increasingly turning to simpler, faster, and more spatially detailed “sketch planning” or “direct demand” models for forecasting rail transit boardings. Planners use these models for preliminary review of corridors and analysis of station-area effects, instead of or prior to four-step regional travel demand models. This paper uses a sketch-planning model based on a multiple regression originally fitted to light-rail ridership data for 268 stations in nine U.S. cities, and applies it predictively to the Phoenix, Arizona light-rail starter line that opened in December, 2008. The independent variables in the regression model include station-specific trip generation and intermodal–access variables as well as system-wide variables measuring network structure, climate, and metropolitan-area factors. Here we compare the predictions we made before and after construction began to pre-construction Valley Metro Rail predictions and to the actual boardings data for the system’s first 6 months of operations. Depending on the assumed number of bus lines at each station, the predicted total weekday ridership ranged from 24,767 to 37,907 compared with the average of 33,698 for the first 6 months, while the correlation of predicted and observed station boardings ranged from r = 0.33 to 0.47. Sports venues, universities, end-of-line stations, and the number of bus lines serving each station appear to account for the major over- and under-predictions at the station level.     

Is accessibility relevant in trip generation? Modelling the interaction between trip generation and accessibility taking into account spatial effects

The influence of accessibility to opportunities in trip generation continues to be debated in the specialised literature given its relevance to simulate phenomena such as induced demand. This article estimates multiple linear regression models (MLR), spatial autoregressive models (SAR), spatial autoregressive models in the error term (SEM) and spatially filtered Poisson regression models (SPO) to discover whether or not accessibility is a significant factor in trip generation using data from the urban area of Santander (Spain). The results obtained provide evidence which shows that, on an intraurban scale, more accessibility to opportunities decreases trip production in private vehicle for work purpose, whereas it increases trip production in other transport modes for non—mandatory purposes. For the correct interpretation of the estimated parameters it was important to consider the direct and indirect effects of the independent variables in the SAR production models. Finally, the validation of the models showed that the SAR and SEM models had a mean squared error slightly lower than the MLR models in predicting overall trip production. This was because the spatial models reduced the correlation of the residuals present in the MLR models. Furthermore, the SPO models performed better in validation mode than all the continuous models.     

Modelling car trip generations for UK residential developments using data from TRICS

Abstract

This paper models trip generation for a cross-section of residential developments around the UK. Consistent with recent literature, the empirical model tests whether trip making patterns for residential developments are independent of car ownership and finds that trip generation is dependent upon car ownership socio-economic factors and site-specific characteristics, in particular land-zone type (e.g. town centre, out of town, etc.). However, public transport services are not found to have a significant relationship with trip generation; consequently, a policy implication of the results is that increasing bus services to residential developments is not associated with a reduction in generated trips.     

Intrapersonal variability and model goodness-of-fit

This paper investigates the effect of day-to-day variability in individuals' travel behavior on the goodness-of-fit of travel demand models estimated with conventional cross-sectional data sets. In particular, this paper examines the effect of day-to-day variability on the goodness-of-fit of least squares regression models of person trip generation. The analytic results show that the conventional R-square goodness-of-fit measure for least squares regression models estimated with cross-sectional data is dependent upon two factors. The first factor is the proportion of the between-person variability that is accounted for by the explanatory variables in the model. The second factor is the proportion of the total variability in the dependent variable that is due to between-person variability. One week activity diary data are used to estimate the relative magnitude of the intrapersonal and interpersonal variability components for a variety of measures of trip-making. This analysis shows that intrapersonal variability may comprise a substantial proportion of the total variability, although the relative magnitude of the intrapersonal variability component varies with trip purpose and across population segments. Generally, however, intrapersonal variability is found to have a considerable effect on the apparent goodness-of-fit of person level trip generation models estimated with cross-sectional data. The paper also illustrates that recognizing the effect of intrapersonal variability on model goodness-of-fit may be useful in interpreting and evaluating model estimation and model transferability results.     

Variability in sound power levels: Implications for static and dynamic traffic models

Classically, one mean vehicle representative of each category is used by both static and dynamic traffic noise prediction models. The spectrum associated with this mean vehicle is determined from a linear statistical regression analysis based on measurement campaigns on a track or in situ. However, the variability of individual vehicle emissions can influence predictions and hinder comparison between static and dynamic models. In order to estimate the induced bias, statistical analysis of the distributions of sound power levels emitted by the individual passage of vehicles during 82 measurement campaigns was carried out. The results show that 92% of the residual regression distributions are Gaussian and that standard deviations can reach 3.6 dBA. The value of the proposed correction term for this case study could reach 1.4 dBA for light vehicles and 1.2 dBA for heavy vehicles. This analysis also shows that the variability in sound power levels and thus the corresponding corrections are higher at the lowest speeds that correspond to urban driving conditions.     

Estimating gravity model parameters: A simplified approach based on the odds ratio

Two computationally simple methods for calibrating the gravity model are presented in this paper. The use of each is demonstrated on several origin-destination trip tables, and the results compared with those obtained from the UTP system gravity model calibration procedure. Our methods, based on the odds ratio, perform at least as well as that of UTP system by several standard criteria.     

A note on the calculation and calibration of doubly constrained trip distribution models

The use of growth factor models for trip distribution has given way in the past to the use of more complex synthetic models. Nevertheless growth factor models are still used, for example in modelling external trips, in small area studies, in input-output analysis, and in category analysis. In this article a particular growth factor model, the Furness, is examined. Its application and functional form are described together with the method of iteration used in its operation. The expected information statistic is described and interpreted and it is shown that the Furness model predicts a trip distribution which, when compared with observed trips, has the minimum expected information subject to origin and destination constraints. An equivalent entropy maximising derivation is described and the two methods compared to show how the Furness iteration can be used in gravity models with specified deterrence functions. A trip distribution model explicitly incorporating information from observed trips, is then derived.It is suggested that if consistency is to be maintained between iteration, calibration, and the derivation of gravity models, then expected information should be used as the calibration statistic to measure goodness of fit. The importance of consistency in this respect is often overlooked.Lastly, the limitations of the models are discussed and it is suggested that it may be better to use the Furness iteration rather than any other, since it is more fully understood. In particular its ease of calculation makes it suitable for use in small models computed by hand.     

Effects of household structure and accessibility on travel

The concept of accessibility has been widely used in the transportation field, commonly to evaluate transportation planning options. The fundamental hypothesis of many studies related to accessibility could be “greater accessibility leads to more travel”. However, several studies have shown inconsistent results given this common hypothesis, finding instead that accessibility is independent of the trip/tour frequency. In addition, empirical aggregate urban modeling applications commonly produce either non-significant or negative (wrong sign) relationships between accessibility and the trip/tour frequency. For this reason, many practitioners rarely incorporate a measure of accessibility into trip/tour generation models out of consideration of the induced demand. In this context, this study examined the effect of accessibility in urban and suburban residences on the maintenance and discretionary activity tour frequencies of the elderly and the non-elderly using household travel survey data collected in the Seoul Metropolitan Area of Korea. The major finding of this study is that a higher density of land use and better quality of transportation service do not always lead to more tours due to the presence of intra-household interactions, trip chaining, and different travel needs by activity type. This finding implies that accessibility-related studies should not unquestioningly accept the common hypothesis when they apply accessibility measures to evaluate their transportation planning options or incorporate them into their trip/tour generation models.     

Traffic assignment by trip type using volume restraint and link restraint for application in small urban areas

This research involved the development of a new traffic assignment model consisting of a set of procedures for an urbanized area with a population of 172,000. Historical, social, and economic data were used as input to conventional trip generation and trip distribution models to produce a trip table for network assignment. This fixed table was divided into three trip types: external-external trips, external-internal trips, and internal-internal trips. The methodology used to develop the new traffic assignment model assigned each of the trip types by varying the diversion of trips from the minimum path. External-external trips were assigned on a minimum path routing and external-internal trips were assigned with a slight diversion from the minimum path. Internal-internal trips were assigned with more diversion than external-internal trips and adjusted by utilizing iterative volume restraint and incremental link restraint. A statistical analysis indicated that assigning trips by trip types using trip diversion and volume and link restraint produces a significant improvement in the accuracy of the assigned traffic volumes.     

Developing an activity-based trip generation model for small/medium size planning agencies

The primary shortcoming of traditional four-step models is that they cannot capture derived travel demand behaviors. However, travel demand modeling (TDM) is an essential input for urban transportation planning. TDM needs to be highly precise and accurate by integrating the accurate base year estimation along with suitable alternatives. Currently, activity-based models (ABMs) have been developed mostly for large metropolitan planning organizations (MPO), whereas smaller/medium-sized MPOs typically lack these models. The main reason for this disparity in ABM development is the complexity of the models and the cost and data requirements needed. We posit however that smaller MPOs could develop ABMs from traditional travel surveys. Therefore, the specific aim of this paper is to develop a probabilistic home-based destination activity trip generation model considering travel time behavior. Results show that the developed model can significantly capture the actual number of trip generations.     

An integrated, dynamic approach to travel demand forecasting

This paper presents a unified approach for improving travel demand models through the application and extension of supernetwork models of multi-dimensional travel choices. Proposed quite some time ago, supernetwork models solved to stochastic user equilibrium can provide a simultaneous solution to trip generation, distribution, mode choice, and assignment that is consistent with disaggregate models and predicts their aggregate effects. The extension to incorporate the time dimension through the use of dynamic equilibrium assignment methods is proposed as an enhancement that is necessary in order to produce realistic models. A variety of theoretical and practical problems are identified whose solution underlies implementation of this approach. Recommended future research includes improved algorithms for stochastic and dynamic equilibrium assignment, new methods for calibrating assignment models, and the use of Geographic Information Systems (GIS) technology for data and model management.     

Four-step travel demand model implementation for estimating traffic volumes on rural low-volume roads in Wyoming

This study develops a four-step travel demand model for estimating traffic volumes for low-volume roads in Wyoming. The study utilizes urban travel behavior parameters and processes modified to reflect the rural and low-volume nature of Wyoming local roads. The methodology disaggregates readily available census block data to create transportation analysis zones adequate for estimating traffic on low-volume rural roads. After building an initial model, the predicted and actual traffic volumes are compared to develop a calibration factor for adjusting trip rates. The adjusted model is verified by comparing estimated and actual traffic volumes for 100 roads. The R-square value from fitting predicted to actual traffic volumes is determined to be 74% whereas the Percent Root Mean Square Error is found to be 50.3%. The prediction accuracy for the four-step travel demand model is found to be better than a regression model developed in a previous study.     

Factors affecting running time on transit routes

This article reports on empirical models of transit mean running time and running time deviation estimated from data collected on transit routes in Cincinnati, Ohio. In general, the models confirm prior hypotheses concerning transit running time. It was found that mean running time is strongly influenced by trip distance, people boarding and alighting, and signalized intersections. Running time deviation at early points on a route propagate as the vehicle proceeds further downstream. The running time deviation model suggests that corrective actions to instability which develops early on a transit route warrants serious consideration. Improvements may also be possible through planning shorter routes. Finally, investment strategies which emphasize improved control rather than modification of existing link characteristics may produce more reliable service. Apart from their explanatory value, the availability of empirical models which relate running time measures to transit route characteristics is seen as being useful for identifying and evaluating strategies to improve service and resource allocation on existing routes and for the introduction of new services. The models also allow for reasonable approximations of running times without the high cost of manual data collection.     

Simplified transport models based on traffic counts

Having accepted the need for the development of simpler and less cumbersome transport demand models, the paper concentrates on one possible line for simplification: estimation of trip matrices from link volume counts. Traffic counts are particularly attractive as a data basis for modelling because of their availability, low cost and nondisruptive character. It is first established that in normal conditions it may be possible to find more than one trip matrix which, when loaded onto a network, reproduces the observed link volumes. The paper then identifies three approaches to reduce this underspecification problem and produce a unique trip matrix consistent with the counts. The first approach consists of assuming that trip-making behaviour can be explained by a gravity model whose parameters can be calibrated from the traffic counts. Several forms of this gravity model have been put forward and they are discussed in Section 3. The second approach uses mathematical programming techniques associated to equilibrium assignment problems to estimate a trip matrix in congested areas. This method can also be supplemented by a special distribution model developed for small areas. The third approach relies on entropy and information theory considerations to estimate the most likely trip matrix consistent with the observed flows. A particular feature of this group is that they can include prior, perhaps outdated, information about the matrix.These three approaches are then compared and their likely areas for application identified. Problems for further research are discussed and finally an assessment is made of the possible role of these models vis-a-vis recent developments in transport planning.     

Individual trip rate transferability analysis based on a decision tree approach

Transferring trip rates to areas without local survey data is a common practice which is typically performed in an ad hoc fashion using household-based cross-classification tables. This paper applies a rule-based decision tree method to develop individual-level trip generation models for eight different trip purposes as defined in the US National Household Travel Survey in addition to daily vehicle miles traveled. For each trip purpose, the models are obtained by finding the best fitted statistical distribution to each of the final decision tree clusters while considering the correlation between the trip rates for other trip purposes. The rule-based models are sensitive to changes in demographics. The performance of the models is then tested and validated in a transferability application to the Phoenix Metropolitan Region. These models can be employed in a disaggregate microsimulation framework to generate trips with different purposes at the individual or household level.