Dynamic Management Tool for Improving Passenger Experience at Transport Interchanges

1. Summary

Thank you very much for taking the time to review this manuscript. Please find detailed responses below and the corresponding revisions/corrections highlighted in blue in the re-submitted manuscript. 

2. Point-by-point response to Comments and Suggestions for Authors

Comments 1: [What are occupancy levels? It is recommended to provide an explanation in the introduction section.]

Response 1: Thank you for your recommendation. We have accordingly included the explanation of the Occupancy Level concept in the Introduction section (lines 59 to 64) and clarified the definition, adding relevant references.  

Comments 2: [In Figure 2, could you please explain the meanings of the labels, letters A to G?]

Response 2: Thank you for pointing it out. We have modified Figure 2 to make it clearer and meaningful. The labels were changed to specify the nomenclature correctly. The new figure is Figure 3 on page 10, and the explanation of its meaning is in lines 337 to 341.  

Comments 3: [For the station's location in this case, a location map can be included to illustrate its position within the transportation network. Similarly, maps can also be utilized to depict the station's entrances and exits, as well as access to the surrounding road network.]

Response 3: Thank you for raising this important point. We have included a location map of the Moncloa Interchange (case of study), including also the Suburban rail and Metro networks, the other interchanges in the system, and the Madrid main transport corridors. Additionally, a detailed map shows the access/egress points for users and buses. Both illustrations are in Figure 2 on page 9, along with a brief description in lines 314 to 324.

Comments 4: [For the model's prediction results, it is recommended to use quantitative indices to discuss the prediction accuracy.]

Response 4: Thank you for your recommendation. In Section 3.4. Model Training and Validation, we define the quantitative prediction indices MAE and RMSE (lines 476 to 485). Then, Section 4.2 Forecasting Level of Service: Predicted Values shows their value in our application and compares the accuracy with other models (lines 550 to 561).

Comments 5: [The coordinates in Figures 4 and 5 are too small to be clearly visible. In Figures 6 and 7, it is recommended to include the units for the vertical axis of the traffic parameters.]

Response 5: Thank you for pointing it out. We have modified the figures to make them clearer. The new figures correspond to Figure 5 and 6 on page 15, Figure 8 on page 17 and Figure 9 on page 18.

1. Summary

Thank you very much for taking the time to review this manuscript. Please find detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the resubmitted manuscript.

2. Point-by-point response to Comments and Suggestions for Authors

Comments 1: [In the Introduction, it would be beneficial to highlight the importance of transfer hubs using international statistics. For example, including data on the volume of transfer terminal usage in countries such as Spain, China, South Korea, and Singapore could enhance the relevance and global appeal of the research, while drawing readers’ interest.]

Response 1: Thank you for your constructive remark. We have included data corresponding to the suggested countries in the Introduction Section, which effectively points out the relevance of these terminals worldwide in the transportation systems. First, we described statistics related to the multimodal trips in big cities – London, Paris, Barcelona, and Nanjing – to highlight the necessity of multimodal hubs. Then, specifically, we pay attention to multimodal trips that include subway trips – Seoul, Singapore, and Madrid – because metro stations are key transfer nodes in the transportation network. This information is included in the first paragraph of Section 1, pages 1 – 2, lines 30-44.  

Comments 2: [The literature review section would benefit from a clearer structure. It is recommended to divide it into three distinct thematic areas, i.e., 1) Multimodal public transport and transfers, 2) LOS evaluation, and 3) Prediction modeling. This structure would improve clarity and allow the reader to better understand the scope and positioning of the study.]

Response 2: Thank you for your valuable recommendation. We have changed the structure of the Literature Review section following your advice. Additionally, we have added some references to support the related concepts and state-of-the-art review further. The changes can be found on page 3, line 99. The new subsections are 2.1. Multimodal Public Transport and Transfers in line 105, 2.2. LOS evaluation in line 188, and 2.3. Prediction Modelling in line 222.

Comments 3: [Additionally, further review of literature specifically focused on public transport transfers, as well as studies that assess the level of service in public transport, would enhance the academic foundation of the paper. These references will help contextualize the current research and underline its novelty within the broader field.: explainable DEA approach for evaluating performance of public transport; Evaluation of transfer efficiency between bus and subway; Assessing vertical transport system installation in subway stations.]

Response 3: Thank you for your helpful observation. We have reinforced the literature review and focused on LOS in section 2.2. LOS Evaluation. We mention the main concepts and add relevant references for the DEA analysis of efficiency and infrastructure assessment. These changes can be found in lines 188 to 221.

Comments 4: [In time-series forecasting, one of the key challenges lies in how to effectively address the lag structure, especially when predicting continuous variables. LSTM models are inherently designed to reflect temporal dependencies, meaning that the current state is strongly influenced by prior time steps. In many cases, this structure leads to high prediction accuracy (e.g., ~90%) simply because the model effectively shifts previous values forward in time. As a result, the predicted output often resembles a right-shifted version of the actual time series, without capturing more complex patterns or deviations. Therefore, the authors should clarify how their model overcomes this persistence effect and demonstrate whether it truly captures underlying dynamics beyond a naive lag-based projection.]

Response 4: Thank you for this insightful comment. We have clarified in section 3.3.1 Data used for modeling that the data sources that our model utilize can reflect part of the complex dynamics of passenger flows in an interchange station (lines 402 to 415, and Table 2 on page 12). Also, we have expanded the explanation of sinusoidal features that the model uses to learn the daily and weekly patterns of the data, by adding Figure 4 on page 12 and description lines 416 to 420.

Comments 5: [The results section should clearly demonstrate whether this pattern has been mitigated. It is recommended to include comparative graphs of observed and predicted values to visually validate the model’s effectiveness.]

Response 5: Thanks again for your valuable recommendation. We have modified section 4.2. Forecasting Level of Service: Predicted Values accordingly. Figures 8 and 9, on pages 17 and 18, now include 24 hours-forward shifted plots of the data known to the model to compare with the prediction of the model and the real values. Also, new paragraphs have been added that elaborate on the superiority of our model over a naive shift model (lines 578 to 624).

Comments 6: [The primary objective of the study is to predict LOS, yet the model is designed to forecast occupancy. The authors should clarify the justification for using occupancy as a proxy and explicitly discuss how the predicted occupancy is reliably converted to LOS levels.]

Response 6: Thank you for your remark. We applied the Level of Services (LOS) thresholds defined in the Highway Capacity Manual (HCM) and the Transit Capacity and Quality of Service Manual to classify the Occupancy Level based on their pedestrian space indicator. The idea that the passengers inside the stations are pedestrians is developed in the manuscript, justifying this metric selection. This is clarified and explained in 2.2. LOS Evaluation section. Additionally, in the Results section, Table 3 relates the values stipulated by the manual, and it corresponds to the real values of the case study, which explains the relationship between both concepts: the occupancy level with a determinate service level, indicating the space used by each traveler.

 Comments 7: [The performance evaluation should include not only MAE, but also other metrics such as RMSE and MAPE, to ensure a more comprehensive assessment of prediction accuracy.]

Response 7: Thank you for pointing this out. Accordingly, we added the RMSE index to the MAE values in a new Table 4 for the validation and test sets. We also include a comparison of RMSE in LSTM with the other models: Last, Linear, Dense and Convolutional in Figure 7.c, on page 16. We have also included some additional explanatory text (lines 558 to 565). We considered using MAPE, but since our target variables are sometimes equal to zero, we have concluded that it is not an adequate measure of accuracy in our case.

Comments 8: [The sentence regarding perceived congestion is not supported by empirical evidence and falls outside the scope of this study. It is recommended to remove this statement and focus on objectively observed outcomes derived from the data.]

Response 8: Regarding your comment, we assume that you are referring to the “perceived service quality” statements that we mentioned twice. We have changed this statement to “offered service quality”, which aligns with our research. This can be found in line 695 in Section 5.1, Data-driven Tool for Dynamic Management of Transport Interchanges.

3. Response to Comments on the Quality of English Language

Point 1: The English could be improved to more clearly express the research.

Response 1: Thank you for your remark. We have identified and corrected the text's typos, as well as grammatical and writing errors.

1. Summary

Thank you very much for taking the time to review this manuscript. Please find detailed responses below and the corresponding revisions/corrections highlighted in blue in the resubmitted manuscript.

2. Point-by-point response to Comments and Suggestions for Authors

Comments 1: [An additional dimension that could enrich the study is a more comprehensive discussion on the integration of transport interchanges within the broader economic and territorial planning framework. Interchanges should not be analyzed solely from an operational or transport-specific perspective, as their real-world impact encompasses a more complex set of functions. In practice, these nodes fulfill three key roles: urban (as structuring elements of the city), transport (as mobility facilitators), and auxiliary or socio-economic (as multifunctional spaces—pôles d’échange—that support commercial, social, and administrative activities). In this regard, I suggest including in the literature review an analysis of the land use and transport interaction, as well as a discussion of the three key functions typically fulfilled by an interchange node. Beyond their well-recognized transport and operational roles, such nodes also perform a socio-economic function, acting as multifunctional hubs (pôles d’échange) that contribute to local development, urban integration, and the enhancement of surrounding areas. Emphasizing this third dimension would provide a more holistic understanding of the strategic importance of interchanges in contemporary urban and regional planning.]

Response 1: We are grateful for your clear and constructive feedback. Your suggestion provides a valuable perspective that has helped us strengthen the Sections: Introduction and Literature Review. We have added new relevant references and further explanations to emphasize the socio-economic significance of terminals not only within the transport network but also in the urban context. We have added the societal dimension to the transport node and space dimensions, already deeply addressed following the Bertolini vision of interchanges. The modifications were targeted in lines 125 to 132 of Section 2.1. Multimodal Public Transport and Transfers. The relevance of the socio-economic aspect in the Interchange is also highlighted throughout the study and considered a condition for selecting the case study, as mentioned in the description in Section 3.1. Case Study for Validation.

Comments 2: [The literature review should include the discussion of Long Short-Term Memory (LSTM) Recurrent Neural Networks, as references to related studies are currently placed in Section 3.3. These works belong in the theoretical background. Section 3.3 should focus exclusively on the implementation details of the LSTM model used in this study.]

Response 2: Thank you for your comment. We have accordingly modified Section 3.3. Phase 2: Short-Term Prediction Model (line 381), limiting this just to mention the selected model. The references about the Long Short-Term Memory were removed for this and included the related information in the Literature Review, specifically in Section 2.3 Prediction Modelling (line 222). The changes can be found from line 222 to 266.

Comments 3: [It is unclear why the publication year appears in bold for some of the references in the bibliography.]

Response 3: Thank you for pointing that out. This was a typo in the reference format, and we have corrected it.