All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
The reviewers appreciated the recent changes to the article so I recommend it for acceptance.
[# PeerJ Staff Note - this decision was reviewed and approved by Claudio Ardagna, a PeerJ Section Editor covering this Section #]
The authors had resolved all issues already.
-
-
According to the authors, the paper presents ten essential strategies for improving ETA accuracy, integrating advanced machine learning techniques, real-time and historical data fusion, and traffic behavior modeling.
The topic is relevant, and the proposal to analyze research in the area is pertinent.
The text is unambiguous.
The abstract is well written and aligns with the paper’s actual scope and tone. While real-world implementations (e.g., Uber, DoorDash, Waze) are discussed as illustrative examples, the authors clarify that the paper offers a thematic synthesis—not an empirical evaluation—of these systems. They have also refined the description of their contributions to avoid overstating the depth of analysis, while still highlighting the relevance of their recommendations for addressing data variability, accuracy-latency tradeoffs, and challenges in dynamic transportation networks.
Although the authors did not use a specific methodology for the systematic review, they present in the introduction the keywords used, the databases searched, as well as the inclusion and exclusion criteria.
The tables effectively present the research findings in a clear and organized manner.
Table 6 provides a summary of bias types in ETA prediction systems, highlighting the main sources of bias, their potential impacts, and corresponding mitigation strategies. This constitutes a significant contribution of the study.
**PeerJ Staff Note:** Please ensure that all review, editorial, and staff comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
Here's a review of the paper for the relevant criteria:
- Clear and unambiguous, professional English used throughout. The language used in the abstract, introduction, and the ten tips appears to be clear, unambiguous, and consistently professional. The writing is formal and uses appropriate terminology for the subject matter of logistics and machine learning.
- Literature references, sufficient field background/context provided. The paper includes a list of references, indicating that the authors have drawn upon existing literature. The introduction provides sufficient field background by highlighting the importance of accurate ETA prediction in modern logistics for operational efficiency, customer satisfaction, and supply chain reliability. It also sets the context by mentioning various applications like last-mile deliveries, freight transportation, ride-hailing services, and e-commerce fulfillment. The challenges in ETA prediction due to real-world factors, urban and rural differences, and the impact of inaccurate ETAs are also discussed, providing further context. The advancements in machine learning for ETA modeling are introduced, establishing the relevance of the field.
- Professional article structure, figures, and tables. Raw data shared. The paper follows a professional article structure with an abstract, introduction, ten distinct tips presented as the main body, and a conclusion. The paper also indicates the presence of figures (Figure 1, Figure 2, Figure 3) and tables (Table 1, Table 2, Table 3, Table 4, Table 5).
- Is the review of broad and cross-disciplinary interest and within the scope of the journal? The review appears to be of broad and cross-disciplinary interest, relevant to researchers in artificial intelligence, transportation modeling, and logistics, as well as professionals in supply chain optimization and urban planning. The abstract and introduction mention the impact of ETA predictions on e-commerce, freight logistics, and ridesharing, indicating their relevance across different sectors. The integration of machine learning techniques, real-time data fusion, and traffic behavior modeling also suggests a cross-disciplinary approach. The paper itself is presented as a review providing actionable insights for researchers and industry professionals.
- Has the field been reviewed recently? If so, is there a good reason for this review (different point of view, accessible to a different audience, etc.)? The authors explicitly state that "The current literature remains fragmented, with diverse methodologies, datasets, and evaluation metrics that make direct comparisons challenging". This provides a clear reason for the review: to "synthesize best practices, identify key challenges, and highlight opportunities for future research and industry adoption". The review aims to consolidate insights from both academic research and industry innovations, offering a structured overview of emerging trends and practical solutions. This suggests that while research exists, a comprehensive and consolidated review like this one is needed due to the fragmented nature of the existing literature and the rapid evolution of the field.
- Does the Introduction adequately introduce the subject and make it clear who the audience is/what the motivation is? Yes, the introduction adequately introduces the subject of ETA prediction by explaining its importance in modern logistics. It clearly defines the scope and mentions various applications. The introduction also makes it clear who the audience is by stating that the review holds particular relevance for researchers in artificial intelligence, transportation modeling, and logistics, as well as professionals in supply chain optimization and urban planning. The motivation for the paper is also clearly stated: the increasing pressure on businesses to provide precise ETAs due to consumer expectations, the negative consequences of inaccurate ETAs (operational inefficiencies, financial losses, diminished customer trust), and the fragmented nature of the existing literature.
- Formal results should include clear definitions of all terms and theorems, and detailed proofs. Based on the provided excerpts, the paper focuses on presenting ten essential strategies for improving ETA predictions. It provides actionable insights, best practices, and identifies challenges and solutions. The content presented in the tips seems to be more descriptive and explanatory, drawing on real-world implementations and research findings.
Yes, the article content appears to be well within the typical Aims and Scope of a computer science journal. The paper extensively discusses machine learning (ML) and artificial intelligence (AI) techniques for improving Estimated Time of Arrival (ETA) predictions. It delves into various computational methods, including deep learning architectures, probabilistic forecasting, reinforcement learning, graph neural networks (GNNs), and spatiotemporal embeddings. The challenges addressed, such as data variability, accuracy-latency tradeoffs, scalability, model interpretability, and algorithmic bias, are significant concerns within the field of computer science, particularly in areas like intelligent systems, algorithms, and data processing. The "Ten quick tips" format suggests an article type that provides a structured overview and practical solutions, which is common in computer science literature to disseminate best practices and emerging trends. The discussion of system design considerations, such as integrating real-time data and handling computational overhead, further aligns with the interests of a computer science audience.
A rigorous investigation appears to have been performed to a high technical and ethical standard from a computer science perspective. The "Survey/Search Methodology" outlines a structured approach using multiple academic databases relevant to computer science (Google Scholar, IEEE Xplore, arXiv) and industry reports. The selection criteria for peer-reviewed papers (2014–2024) featuring empirical validation and real-world case studies indicate a focus on technical rigor. The methodology explicitly mentions efforts to "minimize bias" by balancing academic and industry insights and considering both urban and rural logistics challenges. Furthermore, Tip 10 directly addresses "ADDRESSING ALGORITHMIC BIAS IN ETA SYSTEMS", demonstrating an ethical consideration regarding the impact of computational systems. The assessment of studies for "methodological transparency, dataset reliability, and potential conflicts of interest" also supports the claim of a high technical and ethical standard in the investigation.
The Survey Methodology describes the process of literature collection and analysis with a reasonable level of detail. It specifies the databases used (Google Scholar, IEEE Xplore, arXiv) and the types of industry reports consulted (from leading logistics companies like Uber, DoorDash, Waze). The search strategy involved "broad and targeted terms related to ETA prediction, machine learning in traffic forecasting, and real-time data integration". The inclusion and exclusion criteria for selecting papers are also outlined, focusing on peer-reviewed papers with empirical validation and real-world case studies from 2014 to 2024, while excluding non-English publications, redundant research, and methodologically unclear studies. The analysis prioritized research that compared ETA models, addressed accuracy and scalability, and examined computational trade-offs. While the exact search strings used are not provided, the description offers a good understanding of the methodological approach for a computer science researcher to understand and potentially replicate a similar survey.
The Survey Methodology demonstrates a commitment to a comprehensive and unbiased coverage of the subject within computer science and related domains. The use of multiple relevant academic databases and industry reports suggests a broad search for information. The defined inclusion criteria, focusing on peer-reviewed work with empirical validation in recent years, helps ensure relevance and quality from a computer science perspective. The effort to balance academic and industry insights and consider both urban and rural challenges indicates an attempt to mitigate bias.
However, there are potential limitations. The explicit exclusion of "non-English publications" might lead to the omission of relevant research published in other languages, potentially affecting the comprehensiveness. Additionally, while the search terms are described generally, providing specific keywords or search queries used in each database would enhance transparency and allow for a more precise assessment of the search's comprehensiveness. Details on the number of studies initially identified and the selection process flow (e.g., using a PRISMA diagram or similar) would also provide a clearer picture of the scope and rigor of the survey from a systematic review standpoint, which is increasingly valued in computer science research.
Sources appear to be adequately cited throughout the paper. Each of the ten tips is extensively supported by parenthetical citations to external works [e.g., 87, 100, 105, 115, 120, etc.]. A comprehensive "REFERENCES" section is provided at the end of the paper, listing all the cited sources. The language used in the tips generally synthesizes information from the cited sources, suggesting appropriate paraphrasing rather than extensive direct quotations. This approach is common in computer science survey papers to provide a concise overview of existing research and practices.
Yes, the review is organized logically into coherent paragraphs and subsections. The paper follows a standard academic structure with an abstract, introduction, methodology, ten numbered tips, a conclusion, and references. Each of the ten tips focuses on a specific strategy for improving ETA predictions and is further divided into clear subsections, often using bullet points, to elaborate on different aspects of that tip. The discussion within each tip flows logically, providing explanations, examples often referencing real-world applications by companies like Uber, DoorDash, and Waze, and supporting citations. The use of figures and tables (e.g., comparing base maps, outlining traffic behavior impact, summarizing biases) further enhances the organization and clarity of the information for a computer science audience who often rely on visual aids and structured comparisons. The conclusion effectively summarizes the key strategies and highlights future research directions relevant to the field.
Yes, there is a well-developed and supported argument that meets the goals set out in the Introduction. The Introduction establishes the critical importance of accurate ETA predictions in modern logistics, impacting operational efficiency, customer satisfaction, and supply chain reliability. It outlines the challenges in ETA prediction due to various real-world factors such as traffic patterns, road infrastructure, weather conditions, and stochastic driver behavior. The Introduction also highlights the advancements in machine learning (ML) and artificial intelligence (AI) for ETA modeling while acknowledging persistent challenges like data sparsity and the accuracy-latency trade-off.
The paper explicitly states its goal: "Through the consolidation of insights from both academic research and industry innovations, this paper delivers a structured overview of emerging trends and practical solutions in ETA prediction". The subsequent sections, presented as "Ten quick tips," directly address this goal by providing a structured overview of various strategies for improving ETA predictions. Each tip elaborates on a specific aspect, such as accurate base maps, traffic behavior, driver behavior, real-time adaptability, weather conditions, historical data, unexpected events, user feedback, feature engineering, and addressing algorithmic bias. Each tip includes explanations, real-world examples from companies like Uber, DoorDash, and Waze, and citations to relevant academic literature. This systematic approach demonstrates that the paper effectively develops and supports an argument that meets the goals stated in the Introduction by consolidating insights and providing a structured overview of solutions for ETA prediction.
Yes, the Conclusion does identify unresolved questions, gaps, and future directions. The Conclusion explicitly states that "Future research should focus on improving interpretability and transparency in ML-based ETA predictions through the incorporation of explainable AI techniques and on developing scalable frameworks for global deployment" . It also highlights "the integration of federated learning and privacy-preserving methods" as a direction to further enhance the robustness and trustworthiness of ETA models. These points clearly indicate unresolved questions (how to improve interpretability and scalability), gaps (need for better global frameworks and privacy-preserving methods), and future directions for research in the field of ETA prediction.
The article is generally well-written and offers valuable insights, but there are a few areas that need improvement. The manuscript does not clearly describe the methodology used to select and evaluate the reviewed literature. Including a brief explanation of how sources were gathered, screened, and assessed would strengthen the transparency and reproducibility of the review. Several sections, particularly those discussing advanced machine learning methods, are overly dense with technical jargon. Simplifying the language or briefly defining key terms would make the content more accessible to a broader audience. Although figures and tables are referenced throughout the text, they were not visible in the review version. The final version should ensure that all visuals are properly included, labeled, and high-resolution. The structure of presenting ten separate tips is effective, but the paper would benefit from a summary or synthesis that illustrates how the strategies interconnect or collectively contribute to the development of robust ETA systems. Finally, the manuscript would be strengthened by a short section outlining current limitations in the field and key directions for future research. These improvements would enhance the clarity, rigor, and overall value of the review.
The article would benefit from a clearer explanation of the literature review methodology. Although it references multiple databases and sources, it does not describe how articles were selected, screened, or evaluated for inclusion. Adding details about search terms, selection criteria, and the number of sources reviewed would improve transparency and demonstrate comprehensive coverage. Additionally, some sections are overly technical and dense with jargon, which may limit accessibility for a broader interdisciplinary audience. Simplifying complex terms or providing brief definitions where appropriate could enhance clarity. Lastly, while each of the ten tips is well-developed, the paper lacks a concluding synthesis that connects these strategies into a cohesive framework. Including such a summary, along with a short discussion on current limitations and potential future research directions, would strengthen the overall impact of the article.
The article presents a well-supported argument that aligns with the goals stated in the introduction. Each of the ten strategies is grounded in current research and practical examples, offering a useful synthesis for improving ETA prediction systems. However, the conclusion, while summarizing the key contributions effectively, does not explicitly identify unresolved questions, research gaps, or future directions. Including a brief discussion that highlights open challenges—such as limitations in model generalizability, data sparsity in underserved regions, or the need for benchmark datasets—would provide valuable guidance for future work. This addition would also help demonstrate the replicability and broader impact of the review. Addressing this would enhance the article’s contribution to the literature and better fulfill the expectations for a comprehensive review.
1. The paper reads more like an enthusiastic tech blog than a rigorous literature review—there is no formal methodology to justify the "ten tips" structure. Where is the systematic evidence that these are the ten most critical areas?
2. References are abundant but not critically analyzed. Many are cited to support obvious statements rather than used to evaluate competing models or perspectives. There's a lack of depth in distinguishing between methods.
3. The review is surface-level in several technically demanding sections. Concepts like federated learning, GNNs, and MARL are name-dropped without proper breakdown, giving the illusion of depth.
4. There is no quantitative synthesis. A serious review should include comparative tables, taxonomy diagrams, or performance benchmarking across methods—not just descriptive paragraphs.
5. Industry examples are useful but dominate the academic analysis. The review risks becoming a glorified case commentary unless balanced with critical evaluations of published research.
6. Several sections are bloated with buzzwords—"spatiotemporal embeddings," "context-aware ETA estimation," "probabilistic adaptation"—without clarifying the mechanics. It feels like jargon padding at times.
7. The conclusion lacks intellectual weight. It summarizes well but avoids making bold claims, identifying fundamental unresolved questions, or proposing a future research agenda.
8. The paper avoids hard technical critiques (e.g., failure modes of current models, issues with data bias, reproducibility problems). A more critical stance would make it far more credible.
According to the authors, the paper presents ten essential strategies for improving ETA accuracy, integrating advanced machine learning techniques, real-time and historical data fusion, and traffic behavior modeling.
The topic is relevant, and the proposal to analyze research in the area is pertinent.
The text is unambiguous.
Some points need a review. In the abstract, the authors explained that they analyzed real-world implementations from companies like Uber, DoorDash, and Waze, they provided actionable insights for researchers and industry professionals. And the recommendations obtained address data variability, accuracy-latency tradeoffs, and emerging challenges in dynamic transportation networks, offering a roadmap for optimizing ETA systems.
However, the authors state that they conducted a literature review on several bases. They do not present the methodology used for this survey. What was the methodological strategy for conducting a systematic literature review? Did they use PRISMA, for example?
I suggest that the authors present the methodological approach used, and not only the databases researched. What search strings were used? What were the inclusion and exclusion criteria?
In addition, I also suggest that the authors present comparative tables between the strategies to prove that the ten selected are essential.
.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.