Content uploaded by Apurva Pamidimukkala
Author content
All content in this area was uploaded by Apurva Pamidimukkala on Mar 14, 2024
Content may be subject to copyright.
1
Machine Learning Strategies for Optimizing Urban Parking:
A Comparative Evaluation
Sai Sneha Channamallu,1 Sharareh Kermanshachi, Ph.D., P.E.,2
Jay Michael Rosenberger, Ph.D.,3, Apurva Pamidimukkala, Ph.D.4
1Ph.D. student, Department of Civil Engineering, The University of Texas at Arlington, Arlington,
TX-76019. E-mail: sxc5928@mavs.uta.edu
2Associate Vice Chancellor for Research and Associate Dean of Research, Pennsylvania State
University, State College, PA-16801. E-mail: svk5464@psu.edu
3Professor, Department of Industrial Engineering, The University of Texas at Arlington,
Arlington, TX-76019. E-mail: jrosenbe@uta.edu
4Assistant Professor, Department of Civil Engineering, The University of Texas at Arlington,
Arlington, TX-76019. E-mail: apurva.pamidimukkala@mavs.uta.edu
ABSTRACT
Parking management presents a complex challenge in urban cities, as a scarcity of parking spaces
and the ever-increasing vehicular traffic have led to congestion, environmental pollution, and
overall reduced urban productivity. Addressing the problem requires predicting the exact number
of available parking spaces and categorizing parking occupancy levels. This study aims to achieve
these tasks by employing machine learning models to accurately predict occupancy, thus
optimizing parking resource allocation and enhancing the urban parking experience A dataset
derived from a college campus garage for a period spanning from January 2022 to June 2023 was
used to analyze the performance of various predictive models, including random forest, decision
tree, linear regression, and support vector machine. The models were compared using multiple
evaluation metrics, and the results revealed that the random forest model was the most reliable. Its
strong performance in regression analysis translated into precise estimations of available parking
spaces. Similarly, its capability in classification analysis proved essential for categorizing parking
occupancy into distinct levels, enhancing communication and streamlining decision-making
processes. These findings are significant for improving parking management systems and
contributing to the development of efficient and sustainable parking solutions in urban
environments.
KEYWORDS: Parking Management, Predictive Modeling, Random Forest Model, Decision
Tree, Support Vector Machine, Linear Regression.
INTRODUCTION
Parking management is becoming an increasingly challenging task in urban areas and on college
campuses due to the growing number of vehicles and limited number of parking spaces. The rise
in economic development and urbanization has led to a significant increase in car ownership,
further exacerbating the mismatch between parking supply and demand (Yang et al., 2019;
Pamidimukkala et al., 2023; Patel et al., 2022a). Drivers spend an average of 3.5 to 14 minutes
searching for a parking spot, depending on the location and time of day, which not only leads to
2
frustration and delays, but also wastes fuel and contributes to pollution (Zheng et al., 2015;
Channamallu et al., 2023; Khan et al., 2022a). Minimizing the amount of time spent searching for
parking can help reduce the number of cars circulating in parking lots and decrease traffic
congestion and noise pollution, particularly in the vicinity of parking lot entrances (Caicedo et al.,
2012, Kotb et al., 2017; Etminani-Ghasrodashti et al., 2022).
The inefficient methods employed to search for parking are due to a lack of real-time and
near-future information on the availability of spaces (Yang et al., 2019; Khan et al., 2022b). This
information can be made accessible to drivers by implementing real-time parking maps delivered
through their navigation systems that would provide information about parking availability (Huang
et al., 2018, Sester, 2020; Patel et al., 2022b) and could also be used by parking facility managers
and city planners to better manage and allocate parking spaces based on the predicted demand.
Research has shown that drivers with access to parking space information are 45% more
likely to secure a spot (Caicedo et al., 2006; Khan et al., 2022c), and the realm of influence extends
beyond that of relieving the driver of frustration; it is also crucial for reducing traffic and energy
use (Shoup, 2006; Patel et al., 2023c) and for improving transportation management (Lin et al.,
2017; Etminani-Ghasrodashti et al., 2023). Ignoring parking predictions increases stress and
wastes time and fuel. In the US, drivers spend approximately 17 hours annually looking for a place
to park, costing $345 each (INRIX Research, 2017; Patel et al., 2022a); in the U.K. and Germany,
the amounts of time and costs are even higher. This highlights the importance of addressing the
problem.
Parking occupancy predictions have traditionally been based on static rules and heuristics
that often lack precision and flexibility and rely on manual observations, limited historical data,
and/or basic statistical techniques that may not be accurate or scalable enough for complex urban
parking dynamics (Channamallu et al., 2023a). The development of advanced data-gathering
technologies, availability of extensive parking datasets, and introduction of machine learning
techniques, however, have precipitated a shift towards more data-driven approaches. Utilizing
machine learning models allows parking management systems to better predict parking needs,
optimize space allocation, and provide precise information to drivers, which leads to less
congestion, increased user satisfaction, and greater overall efficiency (Liu et al., 2019; Sun et al.,
2019; Khan et al., 2023a; Patel et al., 2022b).
This study focuses on meeting the challenges presented by the absence of real-time parking
information in urban environments. It centers on analyzing historical occupancy data, time
patterns, and events related to urban parking spaces to develop an intelligent model that accurately
forecasts the availability of parking spaces. The primary objective of the paper is to gain an
understanding of the strengths and limitations of machine learning models by analyzing their
ability to successfully perform both regression and classification tasks within the context of
parking occupancy. The study's goal is to bridge the gap between current parking management
systems and the evolving demands of urban environments by leveraging the predictive power of
machine learning. The results of this research hold practical significance for parking management
systems in urban environments and can aid in creating efficient and sustainable parking solutions.
LITERATURE REVIEW
A significant body of research explores various methodologies for investigating the intricate nature
of parking patterns and providing precise estimations of available parking spaces. It highlights the
effectiveness of various machine learning models, with a particular focus on random forest and
3
decision tree algorithms. Yalcin and Zeydan (2016) discovered that the random forest model is
more accurate than both decision trees and SVR, although decision trees are more computationally
efficient. Wu et al. (2017) explored the suitability of decision tree-based techniques for accurately
predicting parking occupancy and affirmed their effectiveness. Dey and Nath (2019) recognized
the potential of decision tree models for accurately estimating parking availability; Liu et al. (2019)
emphasized the efficacy of gradient boosting decision tree algorithms in forecasting short-term
parking occupancy; and Sun et al. (2019) reported that decision tree algorithms, particularly when
used in conjunction with genetic algorithms, can provide reliable parking availability estimates. A
comparative analysis performed by Farooq et al. (2019) revealed that random forest and decision
tree models consistently achieved higher accuracy and a higher F1-score than SVR and linear
regression models, with random forest also showing superior precision and recall. Anwar et al.
(2019) noted that while SVR demonstrated better recall, linear regression models had the lowest
mean absolute error. He et al. (2020) further corroborated the superior performance of random
forest and decision tree models in terms of accuracy. More recent studies by Srinivasan et al.
(2021), Wu et al. (2020), Kim and Kim (2021), Zhang et al. (2021), and Shi et al. (2021)
consistently validate the greater accuracy of random forest and decision tree models over SVR and
linear regression, affirming their superiority in the field of parking occupancy predictions.
Random forest and decision tree algorithms have consistently demonstrated superior
performance over SVR and linear regression in accuracy and precision (Farooq et al., 2019; He et
al., 2020; Srinivasan et al., 2021). Random forest models are noted for their computational
efficiency while still offering reliable parking availability predictions (Yalcin and Zeydan, 2016;
Wu et al., 2020; Subramanya et al., 2022; Channamallu et al., 2023b), and decision tree models
are also acknowledged for their ability to provide trustworthy forecasts of parking availability (Wu
et al., 2017; Dey and Nath, 2019; Liu et al., 2019; Sun et al., 2019; Khan et al., 2023b; Channamallu
et al., 2023c). This study aims to leverage these insights to further the development of effective
and efficient models for predicting parking occupancy. The collective wisdom from these studies
guides the selection of methodologies, features, and evaluation metrics, paving the way for a
holistic approach to parking management.
METHODOLOGY
The first step of this research entailed an extensive literature review to thoroughly understand the
current state of knowledge in the field. This critical phase laid the groundwork for the study. A
detailed search was performed across various academic databases, including Google Scholar, IEEE
Xplore, Springer, ProQuest, Science Direct, Scopus, and Web of Science, using carefully selected
keyword combinations to precisely target the research topic. The aim of this literature search was
to compile a wide range of relevant scholarly materials, such as conference proceedings, journal
articles, and technical papers, that shed light on the methods and developments in parking
availability prediction. This involved not only finding pertinent publications but also
comprehensively analyzing and integrating the information they provided to form a well-rounded
understanding of the subject. This extensive review informed the subsequent proposal for the
parking occupancy prediction framework depicted in Figure 1.
4
Figure 1 Research framework
Data Collection
The dataset was comprised of 13,104 data points and included the date and hourly occupancy from
January 2022 to June 2023. Figure 2 presents the hourly occupancy in percentages. To determine
historical occupancy rates, a 60-minute lookback window was employed that encompassed various
inputs that provided the model with information regarding occupancy rates 60 minutes before an
attempt was made to make a prediction.
Figure 2 Data set
Variable selection
Relevant features, including the day of the week, time of day, and other contextual information
such as semester, holidays, and special events, were selected based on their potential impact on
parking occupancy. Figure 3 presents the variations in occupancy by the variables employed. The
time of day was revealed to be a crucial factor in predicting parking occupancy as it varied
significantly, for example, between early morning hours and midday. The day of the week was also
shown to be important, as parking demands on weekends and weekdays can be vastly different on
college campuses, with weekdays experiencing higher occupancy due to classes and staff presence.
This study considered the impact of academic semesters on parking occupancy and analyzed the
data separately for fall, spring, and summer semesters, as parking patterns differ across semesters
5
due to variations in class schedules. For example, summer semesters have much lower occupancy
than the fall and spring semesters.
Figure 3 Variations in average hourly occupancy
Data pertaining to holidays was manually gathered from different sources to account for
significantly reduced parking occupancy on these dates that would otherwise be outliers in
predictive models. The analysis also considered semester breaks, recognizing that the parking
occupancy patterns would be different from those of the regular academic sessions, as lower
occupancy is typical during these breaks due to the lack of regular classes and reduced presence
of students. Critical event dates included exams, highly attended sports matches, and concerts, as
such events can lead to significant deviations in normal parking patterns, often causing spikes in
occupancy.
Data Preprocessing
The analysis considered hourly occupancy data as a continuous and categorical variable. For
6
classification purposes, the data was re-categorized into three distinct groups (low, medium, and
high), based on tercile distributions, with specific threshold values set for each category to provide
clearer insights and more accurate predictions. Given the focus on providing accurate occupancy
predictions during typical college operational hours, which are usually from 8 am to 5 pm, the
hourly occupancy data for the hours outside this range were averaged and treated as equivalent to
the 0-hour mark. This adjustment was made to refine the analysis by concentrating on the most
relevant time periods. Data from Saturdays and Sundays was grouped together and categorized as
the weekend to simplify the analysis by treating the weekend as a single entity, acknowledging the
distinct parking patterns typically observed on these days.
Prediction methods
Based on the aforementioned review and relevant information, we selected linear regression,
random forest, SVM, and decision tree models for our research, as they exhibit exceptional
prediction performance and practical feasibility. By comparing the performance of these models,
we aimed to determine their suitability for accurate parking occupancy predictions for a college
campus garage.
Random forest models are particularly notable for their robust handling of complex
variable relationships and efficacy in minimizing overfitting, a common challenge in predictive
modeling. As an ensemble method that amalgamates multiple decision trees, it offers more
generalized and dependable predictions than what might be achieved with individual decision
trees. The decision tree model stands out for its simplicity and ease of interpretation. It is especially
beneficial in contexts where understanding the model's decision-making process is crucial, as it
delineates a clear, straightforward decision pathway based on input features, and it is user-friendly
for stakeholders without a technical background. Linear regression is appreciated for its
straightforward approach in modeling linear relationships between predictors and a target variable
and is particularly effective in scenarios where the linear assumption is valid. It is easy to
implement and understand, and the models are ideal for initial analyses or situations where data
relationships are linear. SVR is lauded for its ability to manage nonlinear relationships, which is
vital for complex scenarios like parking occupancy predictions. Its proficiency in dealing with
nonlinearities makes it a strong candidate for cases where predictor-target relationships are
intricate.
The models were implemented using Python libraries, then were trained and tested by
dividing the dataset into subsets: 70% training and 30% testing. As the input data follows a
sequential nature, such as a time series, we preserved its chronological order during the testing
phase to allow us to evaluate the model's sensitivity to seasonal patterns.
Evaluation metrics
The mean absolute error (MAE), root mean square error (RMSE), and R-squared (R²) were
employed as key metrics in the regression analysis for assessing the performance of predictive
models for parking occupancy. RMSE and MAE are absolute indicators, and comparing their
values helped determine which model had the smallest average prediction error. While RMSE and
MAE focus on the absolute magnitude of errors, R2 captures the proportion of variability that the
model can account for in the target variable. In a classification analysis, precision measures the
proportion of true positives against all positive predictions, recall assesses the model’s ability to
correctly identify all the actual positives, and the F1 Score provides a balance between precision
7
and recall, serving as a harmonic mean of the two.
RESULTS & DISCUSSION
The comparative analysis of the random forest, decision tree, linear regression, and SVR models
revealed their respective strengths and weaknesses in predicting parking occupancy in a college
campus garage.
Regression analysis
The machine learning models were analyzed, using hourly occupancy as a continuous variable,
and the results highlighted the differences in performance and suitability of each model. Figure 4
presents the R2 score of the models before pre-processing the hourly occupancy.
Figure 4 R2 score of models
In the analysis of the models, the random forest model emerged as the standout performer
for predicting hourly parking occupancy, thus highlighting its robustness in capturing the
complexities of parking occupancy data. It exhibited an MAE of 6.26, indicating its high accuracy
in predictions with minimal deviation from actual values; its RMSE of 11.3506 underscored its
precision in predictions, showing minimal variance in errors; and its impressive R² score of 0.7787
signified its ability to explain approximately 78% of the variances in the dependent variable. The
decision tree model was closely aligned with the random forest model and also displayed a
commendable performance. Its slightly higher MAE of 6.3054 indicates strong predictive
accuracy, a comparable RMSE suggests its effectiveness in making precise predictions, and the R²
score of 0.7699 emphasizes its ability to explain a substantial portion of the data's variances.
Conversely, the linear regression model exhibited a weaker performance with a notably
higher MAE of 12.63, which indicated less precise predictions. This was further supported by a
higher RMSE value of 16.038, suggesting greater variance and larger errors in its predictions. The
R² score of 0.558 was lower than that of the tree-based models, indicating that it is less effective
for capturing the variability of parking occupancy data. The SVR model displayed a unique set of
results. Its MAE of 11.1, though lower than that of linear regression, was still higher than the tree-
based models, indicating moderate precision; however, its 18.01 RMSE was higher than all of the
8
models, implying significant variance in its predictive errors. Its R² score of 0.4427 indicates that
it struggles to adequately capture underlying patterns in the data for this application.
Classification analysis
The bar chart presented in Figure 5 shows comparisons of the performances of four machine
learning models (random forest, linear regression, SVM, and decision tree), based on four key
metrics: precision, recall, F1 score, and accuracy.
Figure 5 Results of prediction models
In terms of precision, which measures the proportion of true positive predictions out of all
positive predictions made, random forest achieved a perfect score of 1.00. Linear regression
followed closely behind with a score of 0.99, indicating that it also makes highly precise
predictions. SVM’s score was slightly lower at 0.97, and decision tree’s was the lowest at 0.75,
which suggests that it may generate more false positives than the other models.
All the models performed exceptionally well for recall, which assesses the proportion of actual
positives that are correctly identified. Both random forest and linear regression had perfect recall
scores of 1.00, meaning they correctly identified all the actual positives. SVM also performed well
with a recall of 0.99, and the decision tree's score, although the lowest among the four, was still
high at 0.95.
Accuracy measures the overall correctness of the model, and random forest and linear
regression both scored 0.99, suggesting that they accurately predict the majority of instances. SVM
and decision tree scores were slightly lower at 0.96 and 0.77, respectively, but it should be noted
that although decision tree is less accurate than the others, an accuracy of 0.77 means that it
correctly predicts more than three-quarters of the instances.
The F1 Score is the harmonic mean of precision and recall and provides a single measure of a
test’s accuracy. Random forest and linear regression again led with a perfect score of 1.00, denoting
an excellent balance between precision and recall. SVM’s F1 score of 0.98 was slightly lower but
still indicates a strong performance. The decision tree’s score of 0.74 is respectable but indicates
room for improvement in balancing precision and recall.
To summarize, random forest and linear regression demonstrated superior performance across
all metrics, indicating that they are highly reliable models for the dataset in question. SVM also
9
showed strong performance, particularly in recall and the F1 Score. The decision tree model, while
lagging behind the others, still presents a respectable performance, especially in recall, but may
benefit from further fine tuning to improve its precision and overall accuracy.
CONCLUSION
This paper presents a comprehensive analysis of machine learning techniques for predicting
parking occupancy in a college campus garage. A comparative analysis was conducted of four
models (random forest, decision tree, SVM, and linear regression) by gathering and evaluating
data and assessing the performance and precision of each model. The evaluation utilized metrics
like MAE, RMSE, and R2 for regression analysis, along with accuracy, precision, recall, and the
F1 score for classification analysis.
The in-depth evaluation of both regression and classification models provided substantial
insights. The random forest model’s performance was outstanding, as it achieved perfect or near-
perfect scores in both classification metrics and performance in regression analysis. The linear
regression model showed robust results in classification, but its performance in regression analysis
was not as strong. SVM displayed consistent results in classification but ranked lowest in
regression analysis, and the decision tree model, while having the least impressive classification
metrics, demonstrated a strong performance in regression analysis.
It is clear from integrating these insights that the random forest model's robust performance
in both classification and regression analysis establishes it as the premier choice for predicting
parking occupancy. This dual strength allows for both detailed, number-specific predictions and
categorizations that serve the diverse needs of real-time parking management, enabling users in
making quick assessments and facility managers in effective strategic planning and resource
distribution.
While extreme care was taken in conducting this comprehensive analysis of parking
occupancy predictions, it is important to acknowledge the limitations of the study. This research
focused on a specific urban environment, a parking garage, and the results may not seamlessly
apply to different parking scenarios with unique characteristics. Therefore, applying the random
forest model developed in this study to other contexts should be done with caution. Future research
could expand on these findings by incorporating real-time data, adjusting models dynamically, and
implementing predictive models in practical parking management systems. Further enhancements
in model accuracy and prediction could be achieved through more detailed feature selection,
refining hyperparameters, and adding more contextual variables.
ACKNOWLEDGEMENT
We gratefully acknowledge the support and generosity of the North Central Texas Council of
Governments (NCTCOG), without which the present study could not have been completed.
REFERENCES
Anwar, T., Asadullah, S., Azzam, M., & Abbas, K. (2019). Comparison of different machine
learning models for parking occupancy prediction. In 2019 International Conference on
Frontiers of Information Technology (FIT) (pp. 59-64). IEEE.
10
Caicedo, F., Blazquez, C., & Miranda, P. (2012). Prediction of parking space availability in real
time. Expert Systems with Applications, 39(8), 7281-7290.
Caicedo, F., Robuste, F., & Lopez-Pita, A. (2006). Parking management and modeling of car park
patron behavior in underground facilities. Transportation Research Record: Journal of the
Transportation Research Board, 1956(1), 60-67.
Channamallu, S. S., Kermanshachi, S., & Pamidimukkala, A., 2023. Impact of Autonomous
Vehicles on Traffic Crashes in Comparison with Conventional Vehicles. In International
Conference on Transportation and Development 2023 (pp. 39-50).
Channamallu, S. S., Kermanshachi, S., Rosenberger, J. M., & Pamidimukkala, A., 2023a. A review
of smart parking systems. Transportation Research Procedia, 73, 289-296.
https://doi.org/10.1016/j.trpro.2023.11.920
Channamallu, S. S., Kermanshachi, S., Rosenberger, J. M., & Pamidimukkala, A., 2023b. Parking
occupancy prediction and analysis - a comprehensive study. Transportation Research
Procedia, 73, 297-304. https://doi.org/10.1016/j.trpro.2023.11.921
Channamallu, S. S., Padavala, V. K., Kermanshachi, S., Rosenberger, J. M., & Pamidimukkala, A.,
2023c. Examining parking occupancy prediction models: a comparative analysis.
Transportation Research Procedia, 73, 281-288.
https://doi.org/10.1016/j.trpro.2023.11.919
Dey, K., & Nath, B., 2019. Prediction of parking occupancy using decision tree and support vector
machine. In 2019 3rd International Conference on Computing Methodologies and
Communication (ICCMC) (pp. 131-134). IEEE.
Etminani-Ghasrodashti, R., Hladik, G., Kermanshachi, S., Rosenberger, J. M., Arif Khan, M., &
Foss, A., 2022. Exploring shared travel behavior of university students. Transportation
planning and technology, 46(1), 22-44.
Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J. M., & Foss, A., 2023. Exploring
motivating factors and constraints of using and adoption of shared autonomous vehicles
(SAVs). Transportation Research Interdisciplinary Perspectives, 18, 100794.
Farooq, S., Khaksar, W., Ullah, M. A., & Yaqoob, I., 2019. Comparative analysis of machine
learning algorithms for parking occupancy prediction. Journal of Ambient Intelligence and
Humanized Computing, 10(4), 1343-1355.
He, Q., Li, Z., Huang, C., Zhang, M., & Li, D., 2020. Comparison of machine learning algorithms
for parking occupancy prediction in urban areas. Journal of Advanced Transportation,
2020, 1-12.
Huang, H., G. Gartner, J. Krisp, M. Raubal, and N. Van De Weghe., 2018. “Location Based
Services: Ongoing Evolution and Research Agenda.” Journal of Location Based Services
12 (2): 63–93. doi:10.1080/ 17489725.2018.1508763.
INRIX Research., 2017. Searching for parking costs Americans $73 Billion a year. Retrieved from
http://inrix.com/press-releases/parking-pain-us.
Khan, M. A., Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J. M., Pan, Q., & Foss,
A., 2022a. Do ridesharing transportation services alleviate traffic crashes? A time series
analysis. Traffic injury prevention, 23(6), 333-338.
Khan, M., Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J. M., & Foss, A., 2022b.
Identifying Usage of Shared Autonomous Vehicles (SAVs): Early Findings from a Pilot
Project. In Transportation Research Board 101st Annual Meeting Washington, DC.
Khan, M. A., Etminani-Ghasrodashti, R., Shahmoradi, A., Kermanshachi, S., Rosenberger, J. M.,
& Foss, A., 2022c. Integrating shared autonomous vehicles into existing transportation
11
services: evidence from a paratransit service in Arlington, Texas. International Journal of
Civil Engineering, 20(6), 601-618.
Khan, M.A., Patel, R. K., Pamidimukkala, A., Kermanshachi, S., Rosenberger, J. M., Hladik, G.,
& Foss, A., 2023a. Factors that determine a university community’s satisfaction levels with
public transit services. Frontiers in Built Environment, 9, 1125149.
Khan, M. A., Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J. M., Pan, Q., & Foss,
A., 2023b. Understanding Students’ Satisfaction with University Transportation. In
International Conference on Transportation and Development 2023 (pp. 522-532).
Kim, J. H., & Kim, J. Y., 2021. A comparative analysis of machine learning algorithms for parking
occupancy prediction: Focusing on different time horizons. Sustainability, 13(10), 5557.
Kotb A, Shen Y, and Huang Y., 2017. Smart parking guidance, monitoring and reservations: a
review IEEE Intell. Transp. Syst. Mag. 9(2) pp. 6–16, 2017.
Lin, T., Rivano, H., & Le Mouel, F., 2017. A survey of smart parking solutions. IEEE Transactions
on Intelligent Transportation Systems, 18(12), 3229-3253.
Liu, Z., Li, Y., Liu, Y., & Cai, H., 2019. Parking occupancy prediction based on gradient boosting
decision tree. In 2019 IEEE 2nd International Conference on Information Systems and
Computer Aided Education (ICISCAE) (pp. 358-362). IEEE.
Pamidimukkala, A., Kermanshachi, S., & Patel, R., 2023. An Exploratory Analysis of Crashes
Involving Autonomous Vehicles. In International Conference on Transportation and
Development 2023 (pp. 343-350).
Patel, R. K., Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J. M., & Foss, A., 2022a.
Exploring willingness to use shared autonomous vehicles. International Journal of
Transportation Science and Technology, 12(3), 765-778.
Patel, R. K., Etminani-Ghasrodashti, R., Kermanshachi, S., Rosenberger, J., & Foss, A. 2022b.
How Riders Use Shared Autonomous Vehicles. In Automated People Movers and
Automated Transit Systems, ASCE International Conference on Transportation &
Development, pp. 81-93.
Patel, R. K., Pamidimukkala, A., Kermanshachi, S., & Etminani-Ghasrodashti, R., 2023a. Disaster
preparedness and awareness among university students: a structural equation analysis.
International journal of environmental research and public health, 20(5), 4447.
Patel, R. K., Etminani-Ghasrodashti, R., Kermanshachi, S., Michael Rosenberger, J., & Foss, A.,
2023b. Exploring Factors Affecting Shared Autonomous Vehicles Adoption: A Structural
Equation Modeling Analysis, Transportation Research Board 102 Annual Meeting (pp. 23-
00216).
Patel, R. K., Etminani-Ghasrodashti, R., Kermanshachi, S., Michael Rosenberger, J., & Foss, A.,
2023c. Users’ and Nonusers’ Attitudes and Perceptions of Shared Autonomous Vehicles: A
Case Study in Arlington, Texas. In International Conference on Transportation and
Development 2023 (pp. 241-252).
Sester, M., 2020. “Analysis of Mobility Data – A Focus on Mobile Mapping Systems.” Geo-spatial
Information Science 23 (1): 68–74. doi:10.1080/10095020.2020.1730713.
Shi, W., Chen, X., Zhang, Y., Guo, R., & Xiong, Y., 2021. Comparative analysis of machine
learning algorithms for parking occupancy prediction in smart city. Wireless
Communications and Mobile Computing, 2021, 5551553.
Shoup, D. C., 2006. Cruising for parking. Transport Policy, 13(6), 479-486.
12
Srinivasan, S., Soundararajan, R., Venkata Chalapathy, K., & Balasubramanian, R., 2021.
Comparative analysis of machine learning techniques for parking occupancy prediction.
Journal of Ambient Intelligence and Humanized Computing, 12(5), 5789-5804.
Subramanya, K., Kermanshachi, S., & Patel, R. K., 2022. The Future of Highway and Bridge
Construction: Digital Project Delivery Using Integrated Advanced Technologies. In
International Conference on Transportation and Development 2022 (pp. 14-25).
Sun, Y., Wang, Y., Chen, L., & Gao, Y., 2019. Short-term parking occupancy prediction based on
decision tree and random forest. In Proceedings of the 2019 3rd International Conference
on Industrial and Business Engineering (pp. 135-139). ACM.
Wu, F., Du, Z., Zhang, J., Zhang, S., & Wang, Y., 2017. Parking occupancy prediction based on
decision tree and random forest. In 2017 IEEE International Conference on Real-time
Computing and Robotics (RCAR) (pp. 234-238). IEEE.
Wu, J., Yang, B., & Zhu, D., 2020. A comparative study of machine learning algorithms for parking
occupancy prediction. IEEE Access, 8, 104431-104439.
Yalcin, M. E., & Zeydan, M., 2016. Comparative analysis of machine learning algorithms for
parking occupancy prediction. Expert Systems with Applications, 63, 249-259.
Yang, Y., Sun, S., Qiao, Z., & Zhu, H., 2019. Parking Services in Smart Cities: A Comprehensive
Survey. IEEE Access, 7, 64271-64292.
Yang, S., Ma, W., Pi, X., & Qian, S., 2019a. A deep learning approach to real-time parking
occupancy prediction in transportation networks incorporating multiple spatiotemporal
data sources. Transportation Research Part C: Emerging Technologies, 107, 248–265.
Zhang, K., Zheng, S., Zhao, S., & Li, P., 2021. Comparative study of machine learning algorithms
for parking occupancy prediction: A case study in Beijing. Sustainable Cities and Society,
75, 103287.
Zheng. Y, Rajasegarar. S, Leckie. C., 2015. Parking availability prediction for sensor-enabled car
parks in smart cities, in: Proceedings of the Tenth International Conference on Intelligent
Sensors, Sensor Networks and Information Processing (ISSNIP), IEEE, 2015, pp. 1–6.
