Commute mode choice and number of non-work stops during the commute are joint decisions that have interaction. If an individual chooses a vehicle for the commute, regarding restriction of that vehicle, could has some stops. On the other hand, if an individual need to has some stops, chooses a vehicle for commute regarding number of stops. In this study to consider the interaction between these decisions, we employed copula-based joint modeling framework.The data used in this study is drawn from origin-destination travels data of Shiraz-Iran conducted in 1997. The commute mode choice modeling is undertaken using a multinomial logit model and the number of non-work stops is modeled using an ordered response formulation. To capturing interactive between these decisions several copula functions have been used. Results approve that mode and number of none-work stop choices are interrelated choices by estimating commonly observed factors and dependence parameters with high statistical significance. By determining common effective factors, we can analyze the current situation in the community. also, we can use results for forecasting future travel demand and set some policies leading to promoting trip chaining.

Short-term prediction of traffic variables aims at providing information for travelers before commencing their trips. In this paper, machine learning methods consisting of long short-term memory (LSTM), random forest (RF), support vector machine (SVM), and K-nearest neighbors (KNN) are employed to predict traffic state, categorized into A to C for segments of a rural road network. Since the temporal variation of rural road traffic is irregular, the performance of applied algorithms varies among different time intervals. To find the most precise prediction for each time interval for segments, several ensemble methods, including voting methods and ordinal logit (OL) model, are utilized to ensemble predictions of four machine learning algorithms. The Karaj-Chalus rural road traffic data was used as a case study to show how to implement it. As there are many influential features on traffic state, the genetic algorithm (GA) has been used to identify 25 of 32 features, which are the most influential on models’ fitness. Results show that the OL model as an ensemble learning model outperforms machine learning models, and its accuracy is equal to 80.03 percent. The highest balanced accuracy achieved by OL for predicting traffic states A, B, and C is 89, 73.4, and 58.5 percent, respectively.

There is a relationship between choosing an activity and duration of that activity, especially for non-mandatory ones. Some previous studies have analyzed the decisions about an activity type and duration independently, though some others have used joint models. This paper contributes to the body of knowledge through using Nested-logit and Copula-based models for assessing the existence of interdependency or a hierarchy between non-mandatory activity choice and the relative duration. In the Nested-logit model, it is assumed that error terms of these decisions are interrelated, though one is influenced by another. In contrast, the Copula-based model can accommodate spatial error correlation across observational units without imposing a restrictive distribution assumption on the dependency structures between the error components. The data from Qazvin, a city in Iran, are used for estimating both Nested-logit and Copula-based models and the best variables explaining both choices for each model have been selected. The final models were compared in terms of log-likelihood at convergence and adjusted likelihood ratio index. The results indicated that there are some common influential observed and unobserved factors between these decisions. Also, Copula-based joint model with ρ 0 2 equals to 0.134 outperforms Nested-logit models and provides a better explanatory power.

By predicting and informing the future of traffic through intelligent transportation systems, there is more readiness to avoid traffic congestion. In this study, an ensemble learning process is proposed to predict the hourly traffic flow. First, three base models, including K-nearest neighbors, random forest, and recurrent neural network, are trained. Predictions of base models are given to the XGBoost stacking model and bagged average to determine the final prediction. Two groups of models predict traffic flow of short-term and mid-term future. In mid-term models, predictor features are cyclical temporal features, holidays, and weather conditions. In short-term models, in addition to the mentioned features, the observed traffic flow in the past 3 to 8 hours has been used. The results show that for both short-term and mid-term models, the least prediction error is obtained by the XGBoost model. In mid-term models, the root mean square error of the XGBoost for the Saveh to Tehran direction and Tehran to Saveh direction is 521 and 607 (veh/hr), respectively. For short-term models, these values are decreased to 453 and 386 (veh/hr). This model also brings less prediction error for predicting the first and fourth quartiles of the observed traffic flow as rare events.

No abstract