Bus Arrival Time Prediction Using Machine Learning Approaches
Open accessAbstract
Efficient public transportation systems are critical for urban mobility, requiring precise bus arrival time predictions to enhance service quality and passenger satisfaction. This study thoroughly investigates predictive modeling utilizing machine learning methodologies to forecast bus arrival times along a specific route in Phnom Penh, Cambodia. The dataset contains historical bus arrival data from a specific route, including journey duration, hour of day, day of the week, distance, speed, current bus stop, next bus stop, and weather conditions. The predictive models of this research are built using machine learning methods such as linear regression, XGBoost, support vector machine (SVM), k-nearest neighbors (KNN), and artificial neural network (ANN). These algorithms are used to create prediction models, which are fine-tuned to yield the most accurate estimates of bus arrival times. The usefulness of these models is systematically tested using common performance metrics such as the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE), offering a full assessment of predicting. Extensive experimentation data shows the best performance of XGBoost model, which consistently outperforms other machine learning methods in accurately estimating bus arrival times, with an MAE of 16 seconds, an RMSE of 28.03 seconds, and a MAPE of 2.61%. Our findings contribute to the development of predictive modeling tools in urban transportation planning and management. This study provides a vital tool for improving the performance and sustainability of Phnom Penh’s public transit systems by using the power of machine learning and considering a wide range of parameters, including those not covered by traditional datasets.
AI snapshot
Written from the paper’s full text.
TL;DR
This study compares machine learning models for predicting bus arrival times on Phnom Penh route 11A using historical GPS and weather data. XGBoost achieved the best accuracy, with an MAE of 16 seconds, RMSE of 28.03 seconds, and MAPE of 2.61%.
Key findings
- XGBoost outperformed linear regression, SVM, KNN, and ANN, achieving an MAE of 16 seconds, RMSE of 28.03 seconds, and MAPE of 2.61%.
- ANN with one hidden layer was the best ANN configuration, with MAE 55.12 seconds and RMSE 111.01 seconds; two and three hidden layers had higher errors (MAE 81.31 and 112.42 seconds; RMSE 154.97 and 194.28 seconds, respectively).
- Other tuned models had higher errors: LR (MAE 144.17, RMSE 236.17), SVM (MAE 120.19, RMSE 242.72), and KNN (MAE 109.28, RMSE 166.93). MAPE values were 35.54% for LR, 23.06% for SVM, and 17.85% for KNN.
- The paper concludes that XGBoost and ANN outperform traditional regression and proximity-based methods for bus arrival time prediction.
Methods
- Data: historical bus arrival data for route 11A from Sangkat Pong Tuek to Sangkat Tual Svay Prey II, about 16.4 km with 19 bus stations; GPS tracking collected over two months from July to August 2023, combined with weather data.
- Features included latitude, longitude, date, time, weather, distance, speed, current and next bus stop, hour of day, day of week, and trip duration; the target was arrival time. The dataset was split 80% for training and 20% for testing.
- Models: linear regression, XGBoost, support vector machine, k-nearest neighbors, and artificial neural network; hyperparameters were tuned using GridSearchCV, and performance was evaluated with MAE, RMSE, and MAPE, with k-fold cross-validation also mentioned.
Limitations
- The paper states potential biases in the dataset, such as disproportionate representation of data from specific hours of the day and weather conditions.
- The study uses historical data from a single route (11A) over a two-month period, which the text makes evident and may limit generalizability.
- The paper notes future research will develop IoT sensors for real-time data, indicating the current models are not based on real-time bus data.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 13 cited sources checked · 0 supported · 0 flagged
- Findingreasoning
ANN achieved optimal performance with a learning rate of 0.001, a batch size of 4, and training over 200 epochs.
“ANN achieved optimal performance with a learning rate of 0.001, a batch size of 4, and training over 200 epochs, allowing for gradual learning and precise convergence on complex data patterns.”
Quote in the paperNot a substantial claim - Findingown data
XGBoost's optimal hyperparameters were a learning rate of 0.05, a maximum depth of 5, 300 estimators, a gamma value of 0.1, and a column sample by tree of 1.0.
“XGBoost were found the best of parameters such as a learning rate of 0.05, a maximum depth of 5, 300 estimators, a gamma value of 0.1, and a column sample by tree of 1.0.”
Quote in the paperNot a substantial claim - Findingown data
SVM performed optimally with a regularization parameter C of 100, a gamma value of 0.1, and an epsilon of 0.5.
“SVM performed optimally with a regularization parameter 𝐶 of 100, a gamma value of 0.1, and an epsilon of 0.5, enhancing classification accuracy and margin control.”
Quote in the paperNot a substantial claim - Findingown data
The ANN configuration with one hidden layer achieved the highest predicted accuracy among the ANN configurations tested, with an MAE of 55.12 seconds and an RMSE of 111.01 seconds.
“The findings show that the ANN configuration with one hidden layer delivers the highest predicted accuracy, with an MAE of 55.12 and RMSE of 111.01.”
- Also stated in: § 3.3 Result of Predictive Models · boosted
“Remarkably, Artificial Neural Network exhibits substantial enhancement post-tuning, achieving an impressive MAE of 55 seconds, and an RMSE of 111.01 seconds.”
Quote in the paperNot a substantial claim - Also stated in: § 3.3 Result of Predictive Models · boosted
- Findingown data
KNN showed best results with 7 neighbors, uniform weights, 'auto' algorithm, and a power parameter P of 1 for the Minkowski metric.
“KNN showed best results with 7 neighbors, uniform weights, 'auto' algorithm, and a power parameter 𝑃 of 1 for the Minkowski metric, ensuring effective local pattern recognition.”
Quote in the paperNot a substantial claim
- Interpretationreasoning
Integration of weather conditions into the predictive modeling process enhances forecast precision by accounting for environmental factors that influence transit operations.
“Furthermore, the integration of weather conditions into the predictive modeling process enhances forecast precision by accounting for environmental factors that influence transit operations.”
Quote in the paperSubstantial claim - Interpretationreasoning
XGBoost's accuracy can be attributed to its ability to discern intricate patterns and temporal dependencies within the data.
“The remarkable accuracy achieved by XGBoost can be attributed to its ability to discern intricate patterns and temporal dependencies within the data, surpassing the capabilities of linear regression and distance-based algorithms.”
Quote in the paperSubstantial claim - Interpretationown data
XGBoost outperforms the other machine learning methods tested in accurately estimating bus arrival times.
“Extensive experimentation data shows the best performance of XGBoost model, which consistently outperforms other machine learning methods in accurately estimating bus arrival times, with an MAE of 16 seconds, an RMSE of 28.03 seconds, and a MAPE of 2.61%.”
- Also stated in: § 3.3 Result of Predictive Models · plain
“Of these models, XGBoost has the best alignment between the expected and actual data, which proves more accurate.”
- Also stated in: § 4. CONCLUSIONS · plain
“The study's examination of machine learning models for predicting bus arrival times in Phnom Penh, leveraging historical bus data and weather conditions, underscores the superiority of Artificial Neural Network and XGBoost algorithms over traditional methods.”
Quote in the paperSubstantial claim - Also stated in: § 3.3 Result of Predictive Models · plain
- Interpretationown data
Advanced machine learning algorithms such as XGBoost and ANN outperform classic regression and proximity-based methods in prediction accuracy for bus arrival times.
“These findings demonstrate that advanced machine learning algorithms, such as XGBoost and ANN, outperform classic regression and proximity-based methods in terms of predicting accuracy.”
- Also stated in: § 4. CONCLUSIONS · boosted
“The study's examination of machine learning models for predicting bus arrival times in Phnom Penh, leveraging historical bus data and weather conditions, underscores the superiority of Artificial Neural Network and XGBoost algorithms over traditional methods.”
Quote in the paperSubstantial claim - Also stated in: § 4. CONCLUSIONS · boosted
- Premisecitation
The ANN achieved better prediction accuracy than both the historical data-based model and the regression model.
“The ANN achieved better prediction accuracy than both the historical data-based model and the regression model, which presented a freeway travel time prediction framework that integrated a state-space neural network with preprocessing strategies employing imputation [3,15].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [3]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites [15]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
The k-NN method has the capacity to enhance the accuracy of bus arrival time prediction.
“The k-NN method was formulated, with findings indicating its capacity to enhance the accuracy of bus arrival time prediction [6].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [6]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
An ANN model utilizing arrival time, dwell time, schedule adherence, and distance as primary predictors tends to result in high prediction error rates during adverse traffic conditions due to lack of adaptation to changing traffic conditions.
“The ANN model utilized arrival time, dwell time, schedule adherence, and distance as primary predictors [14], which tends to result in high prediction error rates during adverse traffic conditions due to the model's lack of adaptation to changing traffic conditions [2].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [14]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites [2]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
Support Vector Machine offers clear advantages in addressing challenges such as small sample sizes, nonlinearity, and multivariable classification and regression problems.
“The Support Vector Machine method offers clear advantages in addressing challenges such as small sample sizes, nonlinearity, and multivariable classification and regression problems [9], yet its performance is profoundly influenced by the parameters governing the training process [10,11].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [9]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- May combine two claims. 11
- Classified as probably not a substantial claim. 3
May combine two claims.
Findingown dataXGBoost had the lowest MAPE at 2.61%, while LR, SVM, and KNN had higher MAPE values of 35.54%, 23.06%, and 17.85%, respectively.
“Among the models analyzed, XGBoost demonstrated the lowest MAPE at 2.61%, indicating the highest accuracy in predicting bus arrival times.In contrast, LR, SVM, and KNN models have higher MAPE values of 35.54%, 23.06%, and 17.85%, indicating a poorer predictive precision in this context.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataK-Nearest Neighbors achieved an MAE of 109.28 seconds and an RMSE of 166.93 seconds on the test set.
“| K-Nearest Neighbors (KNN) | 109.28 | 166.93 |”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataLinear regression achieved an MAE of 144.17 seconds and an RMSE of 236.17 seconds on the test set.
“| Linear Regression (LR) | 144.17 | 236.17 |”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataXGBoost achieved an MAE of 16 seconds, an RMSE of 28.03 seconds, and a MAPE of 2.61% on the test set for bus arrival time prediction.
“Extensive experimentation data shows the best performance of XGBoost model, which consistently outperforms other machine learning methods in accurately estimating bus arrival times, with an MAE of 16 seconds, an RMSE of 28.03 seconds, and a MAPE of 2.61%.”
- Also stated in: § 3.3 Result of Predictive Models · boosted
“XGBoost demonstrates significant improvement with an MAE of 16 seconds and an RMSE of 28.03 seconds.”
- Also stated in: § 4. CONCLUSIONS · boosted
“Notably, XGBoost emerges as the standout performer, showcasing exceptional predictive capabilities with minimal errors, including an MAE of 16 seconds, an RMSE of 28.03 seconds, and a MAPE of 2.61%.”
Quote in the paperSubstantial claim- Also stated in: § 3.3 Result of Predictive Models · boosted
May combine two claims.
Findingown dataANN models with 2 and 3 hidden layers have higher MAE values (81.31 and 112.42 seconds) and higher RMSE values (154.97 and 194.28 seconds) than the one-hidden-layer ANN.
“On the other hand, models with 2 and 3 hidden layers have increasing MAE values 81.31 and 112.42, respectively and RMSE values 154.97 and 194.28, respectively, suggesting a diminishing return in accuracy as network complexity increases.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataSupport Vector Machine achieved an MAE of 120.19 seconds and an RMSE of 242.72 seconds on the test set.
“| Support Vector Machine (SVM) | 120.19 | 242.72 |”
Quote in the paperSubstantial claimMay combine two claims.
Study design and conductreasoningThe study employed k-fold cross-validation to ensure the robustness of the models, mitigate overfitting, and provide a more reliable estimate of model performance.
“Moreover, we employed k-fold cross-validation to ensure the robustness of our models, which allowed us to validate the models across different subsets of the data.This approach assists in mitigating overfitting and provides a more reliable estimate of model performance.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenThe dataset has potential biases, such as disproportionate representation of data from specific hour of the day and weather conditions.
“While the models demonstrated varying levels of success in predicting bus arrival times, it is important to acknowledge potential biases in our dataset, such as the disproportionate representation of data from specific hour of the day and weather conditions.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationReal-time transportation information systems enhance travel planning, reduce bus waiting times, and optimize the use of public transport.
“The increasing effectiveness of Intelligent Transport System (ITS) solutions underscores the necessity for a real-time transportation information system, which, by providing commuters with timely information, enhances travel planning, reduces bus waiting times, and optimizes the use of public transport [1].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [1]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
May combine two claims.
PremisecitationThe Kalman Filter model has limited ability to handle non-linear relationships and relies on assumptions that may not hold in complex conditions.
“A study by Yin et al. 2017 considered Kalman Filter model for predicting travel time [8], but we did not consider it in this study due to its limited ability to handle non-linear relationships and its reliance on assumptions that may not hold in complex.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [8]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationSupport Vector Machine performance is profoundly influenced by the parameters governing the training process.
“The Support Vector Machine method offers clear advantages in addressing challenges such as small sample sizes, nonlinearity, and multivariable classification and regression problems [9], yet its performance is profoundly influenced by the parameters governing the training process [10,11].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [10]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites [11]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
May combine two claims.
PremisecitationEnhanced ANNs exhibit accurate performance for predicting both single and multiple stops, with the stop-based ANN preferred for multiple intersections between stops and the link-based ANN better suited for pairs of stops with fewer intersections.
“The reliability analysis demonstrated that enhanced ANNs exhibit accurate performance for predicting both single and multiple stops, with the stop-based ANN being preferred in scenarios involving multiple intersections between stops, while the linkbased ANN is better suited for pairs of stops with fewer intersections [13].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [13]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationThe regression model assumption of independence among different factors is often unrealistic.
“The regression model assumption of independence among different factors is often unrealistic [7].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [7]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
May combine two claims.
PremisecitationThe XGBoost model demonstrates superior performance in terms of both accuracy and efficiency compared to the gradient boosting model.
“The prediction outcomes are contrasted with those of the gradient boosting model, revealing that the XGBoost model demonstrates superior performance in terms of both accuracy and efficiency [12].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [12]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
The paper relies on this without support.
The route 11A data is representative of Phnom Penh's public transit system.
“This study provides a vital tool for improving the performance and sustainability of Phnom Penh's public transit systems by using the power of machine learning and considering a wide range of parameters, including those not covered by traditional datasets.”
The paper relies on this without support.
The two-month data collection period from July to August 2023 is representative of general bus operations on the route.
“GPS tracking was used on city buses to collect data for two months, from July to August 2023.”
The paper relies on this without support.
The models' performance on the test set generalizes to real-world deployment for optimizing bus schedules and improving commuter satisfaction.
“As a result, XGBoost presents a robust solution for optimizing bus schedules, minimizing delays, and improving overall commuter satisfaction.”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Stop 3-4 distance conflicts with the route's cumulative distances
MinorFor trip 1, the reported distance from stop 1 to stop 4 minus the distance from stop 1 to stop 3 is 0.8922, not the 11.549 reported for stop 3 to stop 4.
Reported 11.549 · recomputed 0.8922
“11A_3 11A_4 11.46325 104.82602 11.462506 104.83288 11.549 0.772 1 6 15 Cloudy”
Stop 1-3 duration does not equal its component durations
MinorFor trip 1, stop 1 to stop 2 takes 40 seconds and stop 2 to stop 3 takes 93 seconds. Their sum is 133 seconds, rather than the reported 99 seconds from stop 1 to stop 3.
Reported 99 · recomputed 133
“11A_1 11A_3 11.46453 104.81657 11.46325 104.82602 1.1667 0.707 1 6 15 Cloudy 99”
Cites
XGBoost: A Scalable Tree Boosting System
Chen, T., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939785 · Cited by 54,569 · see citing works
An overview of statistical learning theory
Vapnik, V. N. (1999) · IEEE Transactions on Neural Networks · doi:10.1109/72.788640 · Cited by 6,305 · see citing works
Accurate freeway travel time prediction with state-space neural networks under missing data
van Lint, H., Hoogendoorn, S. P., & van Zuylen, H. J. (2005) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2005.03.001 · Cited by 442 · see citing works
Dynamic Bus Arrival Time Prediction with Artificial Neural Networks
Chien, S., Ding, Y., & Wei, C. (2002) · Journal of Transportation Engineering · doi:10.1061/(asce)0733-947x(2002)128:5(429 · Cited by 362 · see citing works
Bus arrival time prediction at bus stop with multiple routes
Yu, B., Lam, W. H. K., & Tam, M. L. (2011) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2011.01.003 · Cited by 317 · see citing works
Prediction of Pile Bearing Capacity Using XGBoost Algorithm: Modeling and Performance Evaluation
Amjad, M., Ahmad, I., Ahmad, M., Wróblewski, P., Kamiński, P., & Amjad, U. (2022) · Applied Sciences · doi:10.3390/app12042126 · Cited by 240 · see citing works
Bus arrival time prediction using artificial neural network model
Jeong, R., & Rilett, L. R. (2005) · doi:10.1109/itsc.2004.1399041 · Cited by 196 · see citing works
Vanajakshi, L., Subramanian, S. C., & Sivanandan, R. (2009) · IET Intelligent Transport Systems · doi:10.1049/iet-its:20080013 · Cited by 178 · see citing works
Alhakeem, Z. M., Jebur, Y. M., Henedy, S. N., Imran, H., Bernardo, L. F. A., & Hussein, H. M. (2022) · Materials · doi:10.3390/ma15217432 · Cited by 175 · see citing works
Forecasting histogram time series with k-nearest neighbours methods
Arroyo, J., & Maté, C. G. (2008) · International Journal of Forecasting · doi:10.1016/j.ijforecast.2008.07.003 · Cited by 127 · see citing works
Tunnel Surrounding Rock Displacement Prediction Using Support Vector Machine
Yao, B., Yang, C., Yao, J., & Sun, J. (2010) · International Journal of Computational Intelligence Systems · doi:10.1080/18756891.2010.9727746 · Cited by 72 · see citing works
A support vector machine with the tabu search algorithm for freeway incident detection
Yao, B., Hu, P., Zhang, M., & Jin, M. (2014) · International Journal of Applied Mathematics and Computer Science · doi:10.2478/amcs-2014-0030 · Cited by 70 · see citing works
A Freeway Travel Time Prediction Method Based on an XGBoost Model
Chen, Z., & Fan, W. D. (2021) · Sustainability · doi:10.3390/su13158577 · Cited by 69 · see citing works
A prediction model of bus arrival time at stops with multi-routes
Yin, T., Zhong, G., Zhang, J., He, S., & Ran, B. (2017) · Transportation research procedia · doi:10.1016/j.trpro.2017.05.381 · Cited by 50 · see citing works
Bus arrival time prediction at any distance of bus route using deep neural network model
Treethidtaphat, W., Pattara-Atikom, W., & Khaimook, S. (2017) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2017.8317891 · Cited by 48 · see citing works
The smart bus for a smart city — A real-time implementation
Sharad, S., Sivakumar, P. B., & Narayanan, V. A. (2016) · IEEE International Conference on Advanced Networks and Telecommunications Systems · doi:10.1109/ants.2016.7947850 · Cited by 40 · see citing works
Bus Arrival Time Prediction Based on the k-Nearest Neighbor Method
Liu, T., Ma, J., Guan, W., Yue, S., & Niu, H. (2012) · International Joint Conference on Computational Sciences and Optimization (CSO) · doi:10.1109/cso.2012.111 · Cited by 36 · see citing works
Bus Travel Time Prediction: A Comparative Study of Linear and Non-Linear Machine Learning Models
Ashwini, B. P., Sumathi, R., & Sudhira, H. S. (2022) · Journal of Physics Conference Series · doi:10.1088/1742-6596/2161/1/012053 · Cited by 26 · see citing works
Estimation of Bus Arrival Times Using APC Data
Patnaik, J., Chien, S., & Bladikas, A. (2004) · Journal of Public Transportation · doi:10.5038/2375-0901.7.1.1
BUS ARRIVAL TIME PREDICTION USING SUPPORT VECTOR MACHINE WITH GENETIC ALGORITHM
Yang, M., Chen, C., Wang, L., Yan, X., & Zhou, L. (2016) · Neural Network World · doi:10.14311/nnw.2016.26.011
Performance Analysis of Bus Arrival Time Prediction Using Machine Learning Based Ensemble Technique
Gaikwad, N., & Varma, S. (2019) · SSRN Electronic Journal · doi:10.2139/ssrn.3358828
Flagged first, then by citations
Research works printed but not linked: 2
- N. R. Chopde and M. Nichat, "Landmark based shortest path detection by using A* and Haversine formula," Int. J. Innov. Res. Comput. Commun. Eng., vol. 1, no. 2, pp. 298-302, 2013.
- B. Yao, J. Yao, M. Zhang, and L. Yu, "Improved support vector machine regression in multi-step-ahead prediction for rock displacement surrounding a tunnel," Sci. Iran. Trans. Civ. Eng., vol. 21, no. 4, p. 1309, 2014.
Cited by
Still being checked.
- Authors
- Pov, P.
- Affiliations
- Institute of Technology of Cambodia (KH)
- Corresponding author
- Pov, Phannet
- Venue
- Techno-Science Research Journal
- Publisher
- Institute of Technology of Cambodia
- Published
- 2025-04-30
- Volume
- 13
- Issue
- 1
- Pages
- 43
- ISSN
- 3006-4988 (ISSN-L)
- 3006-4996
- OpenAlex
- W7111135293
- Semantic Scholar
- b5d94cd8bb4aaea97d206e7cc613a9b94afea282
- Cited by
- 0
- References
- 11
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Techno-Science Research Journal (publishedVersion)
- Topics
- Physical Sciences › Engineering › Building and Construction › Traffic Prediction and Management Techniques
- Social Sciences › Transportation › Transportation Planning and Optimization
- Physical Sciences › Engineering › Civil and Structural Engineering › Urban Transport Systems Analysis
- Keywords
- bus arrival time prediction, public transport, XGBoost, urban mobility, Phnom Penh, weather conditions, travel time, passenger satisfaction, urban transportation planning
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-11-05.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-11-05.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
