Spatial and Temporal Characteristics of Urban Tourism Travel by Taxi—A Case Study of Shenzhen

Round 1

Reviewer 1 Report

Dear Authors,

your paper is interesting, well done and quite innovative, mainlyfor the subject matter. The aim is clear and the background is well depicted. The design of research is well defined (mainly by fig.2) and the methodology is appropriately described. Results are clearly presented and discussion is adequate. Some lacks in figg. 5 and 6. Bibliography is proper and sufficiently up-to-date.

I propose just few questions/suggestions.

• Please, could you better explain what is the "may day holiday"?
• Please rearrange figg. 5a and 5b (maybe arrange them vertically) and figure 6 (change the scale; try to zoom on the Shenzen district most interesting).
• about KDE, please provide some references.
• fig. 11: please pay attention to the legend (now it is too small)
• Can you better explain the meaning of fig.13?

Author Response

Point 1: Please, could you better explain what is the "may day holiday"?

Response 1: According to your review opinion, we have explained the ‘May Day’ (Line 84 ”In China, ‘May Day’ (a.k.a. International Labor Day) is one of the traditional holidays and the preferred date for tourism in the first half of the year. During this period, Shenzhen, as a coastal tourist city, receives numerous tourists”).

Point 2: Please rearrange figg. 5a and 5b (maybe arrange them vertically) and figure 6 (change the scale; try to zoom on the Shenzen district most interesting). fig. 11: please pay attention to the legend (now it is too small)

Response 2:Following your suggestions, we have reordered the images and redrawn Figure 11, enlarging the legend a little to make it easier to read.

Point 3: about KDE, please provide some references.

Response 3:As you suggested, we have added citations to the English citation address of the KDE algorithm and the source of the kernel method paper in the algorithm.

Point 4: Can you better explain the meaning of fig.13?

Response 4:Thanks for your suggestion, in the first version of the manuscript, we forgot to describe the composition of the network structure. We have now given a concise description of the network structure as well as the visualization method. In line 374 “the nodes indicate the 26 tourist attractions and the edges indicate the flow connection between the attractions. We used the natural break-point method to divide the number of tourist trips into three classes. The red edges have the highest number of trips, the blue edges the second highest and the lowest is the grey edges.”

Author Response File: Author Response.pdf

Reviewer 2 Report

In this article, the authors selected a densely populated mega-city in China with rich tourism resources as a case study. 
First, they extracted tourist trips from taxi trajectories and used kernel density estimation (KDE) to analyze the spatial aggregation characteristics of tourist trip origins. Second, they investigated the spatial dependence of tourist trips using local spatial autocorrelation analysis (i.e., Getis-Ord Gi*). Third, they explored the correlations between the tourist trip origins and urban geographic contextual factors (e.g., catering services, transportation facilities) using a geographically weighted regression model.

Generally speaking, the article is well-organized and structured. There is a smooth connection between the sections.
The title of the article reflects its content and has the appropriate length.
The abstract presents the research gap that this work attempt to address, the main contribution and the research results. The Introduction section gives the general framework and the main contribution of this study. The abstract and introduction are appropriately written.
In the Related Work section, the literature review discusses previous works from different perspectives. 
In Data and Methods, the Study area and dataset and the Methodological framework are analyzed clearly and with many details.
The results are clearly presented and analyzed. The authors made a coherent and focused presentation.
In the Discussion, the authors highlight the key findings and the implications of their study for tourism transportation planning. 

However, some points need to be addressed:
1) The novelty of this paper should be clearly stated, in the Introduction section.
2) The authors miss the experiments setup description. Please, demonstrate the environment of the experiments. 
3) Discussion should be conducted by comparing your work with other ongoing studies in this field. Moreover, the limitations of this study should be elaborated. 
4) Conclusions and future work should be a separate section after the Discussion. Also, it should summarize the outcomes of the research and future directions. It needs a better connection with the research results. Any comparison or limitations of this work should be part of the Discussion.
5) Reference 37 is just a link. I strongly recommend adding it as a footnote rather than in the references section.  Loading the link, it directs to a map in Chinese. It could be better to show it in English. 

To sum up, the authors have written the article in a very elaborate manner and described the details sufficiently.

Author Response

Response to Reviewer 2 Comments

Point 1: The novelty of this paper should be clearly stated, in the Introduction section.

Response 1: Thanks for your suggestion. We have reorganized the Introduction. Taking into account the comments of other reviewers, we have incorporated the Introduction and related studies, and have given details of the novelty of the paper in the final part of the Introduction.

Point 2: The authors miss the experiments setup description. Please, demonstrate the environment of the experiments.

Response 2: Thanks for your suggestion. We have added the experimental setting to section 2.2 (lines 149-150). The experimental environment lists the hardware configuration, the programming language used for data analysis and the tools used during the spatial analysis.

Point 3: Discussion should be conducted by comparing your work with other ongoing studies in this field. Moreover, the limitations of this study should be elaborated.

Response 3: Thanks for your suggestions. We have rewritten the discussion section, and compared other ongoing studies with our results in three aspects: distribution models, movement patterns and modelling of influencing factors. The limitations of this study have been elaborated. The details are provided in section 4 (1), (2) and (3) respectively.

Point 4: Conclusions and future work should be a separate section after the Discussion. Also, it should summarize the outcomes of the research and future directions. It needs a better connection with the research results. Any comparison or limitations of this work should be part of the Discussion.

Response 4: We have also rewritten the conclusion. The results of the study, some interesting findings and future research are highlighted. Meanwhile, we have included the limitations of this study in the discussion section and detailed them in the section 4-(3).

Point 5: Reference 37 is just a link. I strongly recommend adding it as a footnote rather than in the references section.  Loading the link, it directs to a map in Chinese. It could be better to show it in English.

Response 5: Thanks for your suggestions. We have changed reference 37 to a footnote. And the other links in the paper are footnoted, respectively on pages 3-7. We have linked them separately to the English reference address for ease of reading.

Author Response File: Author Response.pdf

Reviewer 3 Report

Type of manuscript: Article
Title: Spatial and Temporal Characteristics of Urban Tourism Travel by Taxi -
A Case Study of Shenzhen

The paper presents a case study on taxi tourism in Shenzhen. For the purpose the authors use a dataset of taxi trajectories (from May 1-3, 2015) that lead to a set of 26 preselected tourist attractions. The analysis combines trip origins, tourist attractions and travel networks and consists of three steps: taxi data preprocessing; spatial analysis with KDE, Getis-Ord Gi* and GWR; tourist travel network analysis.

In my opinion the paper has the following deficiencies:
-Analysis of the related work is rather general and does not discuss technical aspects of the presented methods with relevance to the material presented as the core of the paper.
-The authors only limit their analyses to the selected 26 tourist attractions and do not discuss methods of possible generalization of the approach.
-The authors only present the obtained results and do not give their opinion of the used approach: is it good enough, were the results obtained with the approach satisfactory, are they comparable to other approaches?
-I believe not all of the selected taxi trajectories that ended in the vicinity of tourist attractions were related to tourism. The authors do not consider such possiblity.
-Future work is only signalled.

Additionaly the paper has multiple technical deficiencies:
-There are many language mistakes, e.g. "we choosed Shenzhen..." (line 13), "the pick-up locations of tourists can depict where they probably accommodation" (line 50), "the constant available taxi trajectory" (line 54), "We taken taxi trajectories" (line 72), "It is in a status of continuous development" (line 103).
-Figures are not correctly numbered, e.g. in the text of the paper there are references to figures 3a, 3b, 4a, 4b and the actual figures are numbered only as 3 and 4. Figure 13 is missing a proper legend. 

Author Response

Point 1: Analysis of the related work is rather general and does not discuss technical aspects of the presented methods with relevance to the material presented as the core of the paper.

Response 1: Thanks for your suggestions. Following your suggestions and taking into account the comments of other reviewers, we have included relevant research in the ‘Introduction’ to make the manuscript look more concise. Moreover, our study is concerned with the spatio-temporal phenomena of tourism travel, such as distribution characteristics, network structure characteristics, and influence factors. Techniques in these areas are commonly used in trajectory data, such as complex network metrics calculations and geographical regressions that take into account spatial heterogeneity. And, we focused on the context that is relevant to this study. As tourism transport is an interdisciplinary research direction of tourism and transport, involving the study of tourism flow patterns and the analysis of transport travel patterns, we have focused on these two aspects in order to highlight the research characteristics of this study. And it will be beneficial for researchers in the field of tourism science.

Point 2: The authors only limit their analyses to the selected 26 tourist attractions and do not discuss methods of possible generalization of the approach.

Response 2: Thanks for your suggestions. This is a good suggestion. We have added the ‘Limitations’ section and detailed the limitations of this study in response to your comments. In addition, the attractions are one of the most important spaces in the city for people to relax, travel and keep physically and mentally healthy, and where crowds gather during the tourist season. This is important for the development of the city's tourism resources and the diversion of tourist flows. In some cities, tourism has also become an important part of the urban economy. Finally, based on taxi accessibility satisfaction, our research methodology and the structural characteristics of tourist flows in intra-city attractions can be used to assist in tourism transport planning and tourism corridor improvement. Therefore, we selected 26 tourist attractions.

Point 3: The authors only present the obtained results and do not give their opinion of the used approach: is it good enough, were the results obtained with the approach satisfactory, are they comparable to other approaches?

Response 3: Thanks for your suggestions. Taking into account your comments, we have reworked the discussion section. In the discussion section, we have compared recent studies relevant to this study, including distribution models, travel flow patterns and factors influencing travel. Our results can be corroborated with other relevant studies, some of which, although a reproduction of intuitive impressions, are more objective results than the manual surveys used in traditional tourism research.

Point 4: I believe not all of the selected taxi trajectories that ended in the vicinity of tourist attractions were related to tourism. The authors do not consider such possiblity.

Response 4: Yes, we took this possibility into account when extracting trips. Therefore, in the second step of the method, we started by combining Baidu Street View and Google Remote Sensing images when identifying potential taxi drop-off areas. Baidu Street View allowed us to browse the layout of the roads and facilities around each scenic area, and we identified the most likely drop-off areas after several comparisons. In addition, given that the dates of our study were during the peak tourist season, when there is more congestion around scenic areas and more complex locations for visitors to disembark. We tried to take these circumstances into account when identifying the areas. This will hopefully improve the reliability of the data.

Point 5: Future work is only signalled.

Response 5: Thanks for your suggestions. Considering several suggestions you made earlier, we have taken them into account when writing our future research.

Point 6: Additionaly the paper has multiple technical deficiencies: there are many language mistakes, e.g. "we choosed Shenzhen..." (line 13), "the pick-up locations of tourists can depict where they probably accommodation" (line 50), "the constant available taxi trajectory" (line 54), "We taken taxi trajectories" (line 72), "It is in a status of continuous development" (line 103).Figures are not correctly numbered, e.g. in the text of the paper there are references to figures 3a, 3b, 4a, 4b and the actual figures are numbered only as 3 and 4. Figure 13 is missing a proper legend.

Response 6: Following your advice, we have carefully checked the relevant technical details. And the manuscript was submitted to a specialised English language agency for checking. We will attach the relevant language certificates.

Author Response File: Author Response.pdf

Reviewer 4 Report

Refer to the attached file, please.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 4 Comments

Point 1: The characteristics of tourism should be analyzed at the very least based on two different day groups (d) (e.g., weekdays and weekends) according to different seasons (s). And then some differences and findings from the analysis results (for d*s cases) should be discussed.

Response 1: Thanks for your suggestions. We had the same idea when conducting this study. We had the same idea in mind when conducting this study. However, due to the difficulty of obtaining data and the objectives of the study, we only focused on the characteristics of tourist travel behaviour during May Day. Your suggestion is also an objective for our future research and is included in section 5 (Page 16, Line 516-517).

Additionally, taxis are the preferred mode of travel in many countries, particularly for individual journeys. Existing studies on travel behaviour using taxi trajectory data have focused on the characteristics of commuters. Few studies have focused on the urban transport dynamics in a destination from a tourism perspective. For this reason, this study chose a date with a high volume of tourist travel - May Day. In China, this is a preferred time for tourism in the first half of the year. Other periods, such as the National Day and the Dragon Boat Festival, have the same travel fervour. However, unlike everyday travel studies, such travel times in a year are uncommon and the collection of travel data is more difficult. Although, there are many studies on urban transport, whether the same characteristics are present when travelling during peak tourist periods has not been adequately studied. This is one of the main research objectives of this study. We believe that this study will be informative and useful to scholars in the tourism. 

Point 2: I don’t think it is sufficient to use only three-day data in order to ensure the reliability of this results. Based on my experience on vehicle trajectory big data, at least three-month trajectory data can include some representative characteristics of complex trips in the behavior of transportation network.

Response 2: Thanks for your suggestions. The results of the three-day data volume study do suffer from a representativeness flaw. However, some of the findings in the study are still useful for tourism product development, passenger flow regulation and guidance. The lack of data and representativeness issues are also summarized in section 4(3) (Page 15, Line 477).

Point 3: The findings are too common and old-fashioned. So I think the findings can be connected the analysis results of review opinion 1.

Response 3: According to your review opinion, we have rewritten section 4 and compared the discussion with recent relevant studies, e.g. on distribution models, movement patterns and influencing factors. In the conclusions section, some of the findings are similar to everyday traffic studies and previous impressions. However, the results obtained using the taxi data approach are more credible than those from traditional manual surveys of tourist traffic. The conclusions drawn from the data also confirm the impressions for tourism management.

Point 4: I think this manuscript has many unnecessary parts. So I think it can be shortened without any unnecessary parts for anonymous readers.

Response 4: Thanks for your suggestions. According to your review opinion, We have included related research in the Introduction section and removed some discussion. Meanwhile, in the research methods section, we have removed some details of familiar algorithmic models and given specific reference websites that we used.

Point 5: Some Figures should be modified. For example, time(hr), and length(km) in Fig 4.

Response 5: Thanks for your suggestions. We have modified all the pictures in the manuscript to add the units.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

My remarks were taken into account by the authors in the revised version of the paper.

At line 210 an expression for the GWR model is missing.

Author Response

We have modified this part. As GWR is a commonly used formula and links are given in the previous sections, so we have removed line 210.

All authors thank you for your patience in reviewing the manuscript and for your helpful comments. We will check the manuscript carefully before submitting the final draft again

Author Response File: Author Response.pdf

Reviewer 4 Report

I think this manuscript was improved to be suitable for the publication of this journal.

Author Response

All authors thank you for your patience in reviewing the manuscript and for your helpful comments. We will check the manuscript carefully before submitting the final draft again

Author Response File: Author Response.pdf