This paper presents a study in capturing the impacts of the mandatory pandemic-induced telework practice on workers’ perceptions of the benefits, challenges, and difficulties associated with telecommuting and how those might influence their preference for telework in the future. Data was collected through an online survey conducted in South Florida in May 2020. Survey data showed that telework indices (either measured through actual behavior or stated preference) before, during, and after the pandemic were heterogeneous across socio-economic, demographic, and attitudinal segments. Before the outbreak, males, full-time students, those with PhD degrees, and high-income people showed higher percentages of involvement in jobs with a telework option. They also had higher pro-technology, pro-online education, workaholic, and pro-telework attitudes. During the pandemic, professional/managerial/technical jobs as well as jobs with lower physical-proximity measures showed the highest telework frequency. In view of future telework preferences, our analysis showed that those who were more pro-telework, pro-technology, and showed less dislike of telework dislike preferred higher telework frequency. A structural equation model was developed to assess the impacts of different predictors on telework behavior before the pandemic and preferences after the pandemic. While telework frequency before the pandemic was highly affected by the pro-telework attitude, the after-pandemic preferences were influenced by several other attitudes such as dislike telework, enjoy interaction, workaholic, as well as productivity factors. This might confirm the assumption that the mandatory practice through the pandemic has provided employees more experiences with work-from-home arrangements, which could reshape decisions and expectations around telework adoption in the future.

To gain a better understanding of online education status during and after the pandemic outbreak, this paper analyzed the data from a recent survey conducted in the state of Florida in May 2020. In particular, we focused on college students’ perception of productivity changes, benefits, challenges, and their overall preference for the future of online education. Our initial exploratory analysis showed that in most cases, students were not fully satisfied with the quality of the online education, and the majority of them suffered a plummet in their productivities. Despite the challenges, around 61% believed that they would prefer more frequent participation in online programs in the future (compared to the normal conditions before the pandemic). A structural equation model was developed to identify and assess the factors that contribute to their productivity and future preferences. The results showed that lack of sufficient communication with other students/ instructor as well as lack of required technology infrastructure significantly reduced students’ productivity. On the other hand, productivity was positively affected by perceived benefits such as flexibility and better time management. In addition, productivity played a mediating role for a number of socio-economic, demographic, and attitudinal attributes: including gender, income, technology attitudes, and home environment conflicts. Accordingly, females, high income groups, and those with home environment conflicts experienced lower productivity, which indirectly discouraged their preference for future online education. As expected, a latent pro-online education attitude increased both the productivity and the future online-education preference. Last but not the least, Gen-Xers were more likely to adopt online-education in the post pandemic conditions compared to their peers.

This paper presents the results of an analysis focusing on large truck-involved work zone fatal crashes using seven-year crash data in the State of Florida. Decision tree/random forest models were applied to specifically detect critical crash patterns that result in a fatality outcome. Because of the imbalanced nature of crash severity data (very low frequency of fatal crashes compared with property damage only or injury), data were treated using random and systematic over-sampling techniques. Marginal effects were addressed using Shapley values to increase model explainability. From a methodological perspective, results showed that the combination of over-sampling techniques with ensemble random forests could significantly improve model performance in predicting fatal crashes (compared with conventional logistic regression models). Primary contributors included pedestrian involvement, lighting conditions, safety equipment, driver condition, driver age, and work zone locations. For pedestrian crashes, factors such as dark-not lighted conditions, distracted truck driver, and driver’s age (young drivers outside city limits, senior drivers inside city limits) were highly likely to be fatal. For non-pedestrian crashes, the combination of front airbag deployment with any restraint system other than shoulder and belt was quite likely to be fatal. Also, abnormal driver conditions increased the risk of a fatal outcome. Additionally, the presence of female drivers (as the second driver in multiple vehicle crashes) highly decreased crash severity, probably because females typically drive more carefully than males. Interestingly, truck driver actions and maneuvers as well as roadway design and other physical environment features (i.e., number of lanes, median type, roadway grade, and alignment) did not show significant contribution to the model.

It is well-known that the appearance of measurement errors spoils the traditional results in survey sampling. The concept of correlated measurement errors (CMEs) is true in various practical situations, but this has been seldom considered by researchers in survey sampling. In this article, the influence of the CME under simple random sampling (SRS) has been considered over some prominent classes of estimators for the population mean. The first-order approximated formulae of the mean square error of the introduced estimators are reported, and a comparative analysis has also been conducted with traditional estimators. The theoretical findings are extended by a broad spectrum computational study using real and artificial data.

This paper presents an investigation of the use and frequency of use of ride-hail services. In particular, we explored the role of generational effects and the heterogeneity involved in Millennials’ decision making when it comes to ride-hail choices. Using an ordered logistic regression structure, different statistical models were developed and tested, including fixed-effects and random parameter models, as well as the inclusion of interaction effects and attitudinal factors. Initial results from the fixed-effects model showed that the younger cohorts, including Millennials and Generation Z, showed a significantly positive preference for more frequent ride-hail use, whereas the older cohorts’ preferences (Generation X, Baby Boomers, and older) did not show any significant effects on ride-hail frequency. In the next step, the presence of heterogeneity among Millennials was tested using random parameters. The results confirmed that Millennials’ usage of ride-hail was heterogeneous, and this was statistically significant at the 90% confidence interval $({\sigma }^2=0.61,\rho =0.097)$ . To identify sources of heterogeneity, interaction effects were added to the model. Accordingly, use of ride-hail was more popular among middle-aged Millennials (30 to 34 years old) and Millennials with higher incomes. Likewise, attitudes such as cost sensitivity (toward private vehicle ownership), and being a rational user resulted in higher frequency ride-hail use across Millennials. On the contrary, unemployed Millennials were less likely to utilize ride-hail. The results from this study provide a more transparent picture of current ride-hail market segmentation, which could help predict the future market comprising autonomous vehicles and other emerging mobility options.

Autoencoders have been extensively used in the development of recent anomaly detection techniques. The premise of their application is based on the notion that after training the autoencoder on normal training data, anomalous inputs will exhibit a significant reconstruction error. Consequently, this enables a clear differentiation between normal and anomalous samples. In practice, however, it is observed that autoencoders can generalize beyond the normal class and achieve a small reconstruction error on some of the anomalous samples. To improve the performance, various techniques propose additional components and more sophisticated training procedures. In this work, we propose a remarkably straightforward alternative: instead of adding neural network components, involved computations, and cumbersome training, we complement the reconstruction loss with a computationally light term that regulates the norm of representations in the latent space. The simplicity of our approach minimizes the requirement for hyperparameter tuning and customization for new applications which, paired with its permissive data modality constraint, enhances the potential for successful adoption across a broad range of applications. We test the method on various visual and tabular benchmarks and demonstrate that the technique matches and frequently outperforms more complex alternatives. We further demonstrate that implementing this idea in the context of state-of-the-art methods can further improve their performance. We also provide a theoretical analysis and numerical simulations that help demonstrate the underlying process that unfolds during training and how it helps with anomaly detection. This mitigates the black-box nature of autoencoder-based anomaly detection algorithms and offers an avenue for further investigation of advantages, fail cases, and potential new directions.