Commuting Analysis of the Budapest Metropolitan Area Using Mobile Network Data

Round 1

Reviewer 1 Report

  -- SUMMARY --   In all, this is a well presented paper that has the potential to be a good publication - there are three key issues to address here in my eyes:

1. The authors need to clearly articulate why this work is important: what is novel here, and what is the gap in knowledge that it seeks to fill? This should be clearly articulated in the intro/lit review and the conclusion.
2. The authors need to present more detail in both the methods and results in much more detail.
3. A more thorough treatment of the limitations (including ethical and practical) of this work is required.

  -- MAIN POINTS --   The literature review reads rather like a list of methods that have previously been applied, rather than a critical analysis of the development of the field and the current 'state of the art'. Crucially, this does nothing to lead the reader to the gap in knowledge that your proposal seeks to fill. You should re-frame the introduction (minor edit) and literature review (significant re-working) to fulfil this purpose - this will add great value to the manuscript, as without a clear articulation of why this work is needed, the purpose of this research is unclear (especially as even a cursory search of Google Scholar demonstrates lots of similar work having been undertaken over the past decade - particularly around the Orange 'D4D' competitions, which are a notable absence from this review).   The method is extremely brief and high level (i.e. lacking in detail of how each step was achieved). It would greatly facilitate reproducibility (and likely the citation rater of this publication) if you were a little more specific about how you undertook this analysis - what exactly did you do, and with what software? Where scripts were created to facilitate analysis - could these be shared in an open repository such as GitHub, GitLab or similar? You also do not mention the statistical methods that you later use in your results (though they are relatively standard - they should still be explained and justified - including ensuring that your data meet the assumptions for these statistical tests).   The limitations section is extremely brief - and needs a more thorough treatment - in particular the assumptions that underlie the analysis (including that Vodafone subscribers are distributed both spatially and demographically in order to be representative of the population; and the temporal differences between the datasets) should be described. There is also a significant omission here, which is around the ethics of using this type of data source - if it is possible to determine the home and workplace location for an individual, then it would be simple to identify individuals, meaning that the data are not anonymous (even if the records do not directly identify them). This requires significant attention to ethical responsibilities of the researcher, which should be explained in this paper. Discussions around similar issues could be identified in papers such as: L. Taylor (2016) No place to hide? The ethics and analytics of tracking mobility using mobile phone data. Environment and Planning D: Society and Space 34 2 319-336 (NB: the reviewer has no relationship to this manuscript or its author).   Finally, the conclusion provides a brief summary of the findings (that the CDR analysis produces similar patterns to the census analysis), but it falls short of articulating why this matters - what is the novelty here? What does this add to existing knowledge? The broad approach is not novel, I don't think, but is there something specific about commuting that has not been previously discussed? Is it simply a case study in a different context to previous work? Being clear on this would greatly increase the value of this manuscript.   -- MINOR POINTS --  

• p1 l21: lived > live
• p1 para 1: it is not clear how these geographies relate t each other - a context map would be very useful (e.g. what is Pest County and the "agglomeration of Budapest", in comparison with the overall "Budapest")? I would re-phrase this to make it clearer and add an illustrative map. From context - I would assume that the "agglomeration" is something akin to a Central Business District?
• p2 l41: what does "its methodology should be established" mean? I would re-phrase. Does this refer to the need for a consistent, 'standard' method for analysis?
• p2 l44: define CDR at first use (Call Detail Record)
• Figure 1 - move down to the Data section
• p4 l136: the information about the validation process ("So, the settlement and — in the case of Budapest — district-based population data [2] is applied from the HCSO.") is quite vague - can you elaborate on exactly what you did here please?
• p4 l144: should "cell" be "cell tower" or similar?
• p4 l154+: as above, these results presume a knowledge of Budapest on the part of the reader - which will limit your potential audience significantly. I would strongly recommend an illustrative map demonstrating the Buda and Pest regions of the city and the Buda Hills, as well as the features in the above suggestion. (see also places referenced on p7)
• Figure 3 is a little mis-represented, the CDR data is the estimated number of Vodafone Subscribers, not the estimated population (hence the very different scale on the colour bar) - I would at least correct the explanation, and perhaps consider either scaling the CDR data using the market share in order to create an estimated population value or simply normalising both to a 0-1 scale to facilitate comparison.
• Figure 5 requires a qualitative colour scheme for the districts (currently you are using a two-colour ramp) and the district colours should not match with the scheme used for the network edges

 

Author Response

We would like to thank you for your valuable comments. The detailed answers are in the attached PDF.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper showcases an approach to estimating commuting flows in Budapest, which utilizes mobile phone data. The paper might be improved in the following ways.

1. The key contributions should be highlighted in relation to the existing studies. Right now, it's unclear whether the main contribution of the study is about methodology or the specific findings about Budapest.

2. One way to address the above issue is to improve (re-structure) the literature review section. The literature review sections as it stands now looks like an assembly of findings from different studies without a central theme(s). It should be re-structured / re-organized so that it becomes clear what is the gap you are trying to filling in this study and whether that gap was about methodology or about Budapest.

3. The paper mentions multiple times "agglomeration of Budapest" and "settlements outside the agglomeration", it would be helpful to show a map (perhaps in the introduction section) illustrating the different parts of the study area.

4. Acronyms such as "CDR" should be explained in the text (not just in the end note). 

5. In Figure 1, it would be helpful to clarify what each category indicate in the juxtaposed figures. Maybe it'd be less confusing to replace the central column of text with a legend explaining what each color indicates. For example, the dark green color should indicate the category "SIM card with more than 1000 records", right?

6. It would be good to have more details regarding the mobile phone data. For example, what are the "activity records"? What do they look like in the raw data? How are the "cells" determined? What is their size? How to determine a device is present in a certain cell? And how to count the frequency of presence of devices?

7. In the results and discussion section, specific districts are mentioned (e.g., "Districts 16 and 17"), but it is not very informative because they are not labeled/marked on the map.

8. In Figure 4 (or in the text) clarify what each point represents - what is the unit of analysis? If each point represent the population of a specific district, then how are the districts defined and how many are they in Budapest?

9. In Figure 6, it mentions three settlements but only shows two (where are 'g' and 'h'?) Also, the right column of figures seem to show the working locations with (instead of without) local workers, right?

10. In section 5.6 (limitations), the potential bias of using mobile phone data (especially from only one mobile phone operator) should be discussed.

Author Response

We would like to thank you for your valuable comments. The detailed answers are in the attached PDF.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear authors,

Please find in the attached file my comments.

Thank you.

Comments for author File: Comments.pdf

Author Response

We would like to thank you for your valuable comments. The detailed answers are in the attached PDF.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

In all, this is a substantial improvement. Though issues still remain, these are probably not significant enough to prevent publication.   To address the main points from my previous review:   The literature review: The authors have clearly added a lot of new material to the literature review, which is great - this has certainly made the section a 'better list' - but does not quite go as far as to be articulate the gap in the literature that this work fills, nor provide any critical analysis of the literature. Both of these would improve the paper.   The description of the data method and are much better now, especially with the inclusion of the Jupyter Notepads.   The limitations section is also greatly improved.   The conclusion: this has not been substantially edited, I still this that this could be improved, but the current version is probably sufficient now that the rest of the paper is improved.

 

Author Response

Two new paragraphs are added to the literature review to cover this issue. See lines 158-171.

The conclusion is also expanded with a new paragraph (see lines 559-565) based on Reviewer 3's remarks.

Reviewer 3 Report

Dear authors,

Please find in the attached file my comments to this revised version of your manuscript. Thank you.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

The detailed response is in the attached PDF.

 

Author Response File: Author Response.pdf