Extending Geodemographics Using Data Primitives: A Review and a Methodological Proposal
Round 1
Reviewer 1 Report
To the authors of "Extending geodemographics using data primitives: a review and a methodological proposal"
Introduction is short, what is good, but may be, lacks of mentionning the objectives and applications domains of urban geodemography (transportation, marketing, welfare policies ...) which may have differing purposes and may promote different ways of data collection and uses.
section 2 provides a review of geodemographic classification approaches, which looks relevant and gives a picture of the current baseline. Section 3 develops the current limitations.
§3.1: temporal dynamics: you could mention what technologies are used for measuring change (eg. from simple difference to Markov chain or deep learning), which may have differing sensitivity to spatial and temporal scale issues.
§3.2: hard versus soft approaches: OK: here you mention fuzzy versus crisp alternative.
§3.3:
211-213: Yes, new opportunities are there. Could you just provide examples of data that are updated with greater frequency.
§4.1:
248-253: data primitives: try to detail how they differ from mere general parameters or indicators or "proxy variables", named at line 292. You mention 'approx spaces' or 'quantified conceptual overlaps', but it doesn't really enlighten the concept, if not explained in more detail. Possibly give an example from the land-use domain..
255-256: orthogonal: do you mean weakly correlated? or qualitatively uncorrelated?
section 4.2:
line294: comment: could you use Table 1(continued) instead of Table 2?
Tables1 and 2: interesting choices.
You should mention somewhere that not every data primitive (though quantified) is not of the same kind: example: additivity or comparability is ok with house price (if you mean average per square_foot) but not easy with income inequatity (Gini?); also: Professional occupation versus Low-skilled occupation versus Unemployment.
Also to mention: space and time resolution for each data primitive is important. Example: Population density and Population flux (update frequency is crucial).
§4.3: Analysing State and Change
321-323: you say "at appropriate spatial and temporal resolutions". That's the point! You should develop a bit in relation with the different kind of data primitives, as mentioned in the previous remark.
333-337: by using "vectors of change" do you mean they are attached to a pixel representation (data cube) rather than a polygon representation?
§4.4. Problems yet to be solved:
Ok: several above remarks could be addressed here, after having been mentioned earlier in your text. That section deserves some more details about the issues: possibly with reference to similar issues from land-cover ones.
Review conclusion:
interesting paper, a critical review and a proposed perspective (vector of changes based on data primitives). However a few improvements are probably necessary before final approval for publication. What requires a "major revision".
Ref: you may consider adding in Section 2:
Douglas S. Massey and Nancy A. Denton. The Dimensions of Residential Segregation. Social Forces, Vol. 67, No. 2 (Dec., 1988), pp. 281-315.
The Dimensions of Residential Segregation (urbanpolicy.net)
Author Response
ijgi-1224200: Extending geodemographics using data primitives: a review and a methodological proposal
Reviewer 1
To the authors of "Extending geodemographics using data primitives: a review and a methodological proposal"
Introduction is short, what is good, but may be, lacks of mentionning the objectives and applications domains of urban geodemography (transportation, marketing, welfare policies ...) which may have differing purposes and may promote different ways of data collection and uses.
Response: This has been done in the opening sentence, lines 23 to 28, in the revised paper.
section 2 provides a review of geodemographic classification approaches, which looks relevant and gives a picture of the current baseline. Section 3 develops the current limitations.
- 3.1: temporal dynamics: you could mention what technologies are used for measuring change (eg. from simple difference to Markov chain or deep learning), which may have differing sensitivity to spatial and temporal scale issues.
Response: This has NOT been done.
This section is not concerned with temporal methods per se but with i) the data that underly such approaches, and ii) the nature of change when dealing with classifications. I am concerned that if some of the methods commonly used in temporal analysis are described here then this will detract from the main points. These are that the data used to support geodemographics are not well suited to change / temporal dynamics AND that the concept of change in the context of classification, misses many important smaller signals of change, that may be important despite being too small to result in a class label change.
- 3.2: hard versus soft approaches: OK: here you mention fuzzy versus crisp alternative.
- 3.3:
211-213: Yes, new opportunities are there. Could you just provide examples of data that are updated with greater frequency.
Response: This has been done. Lines 218-224 indicate some of the higher temporal resolution data that are available.
- 4.1:
248-253: data primitives: try to detail how they differ from mere general parameters or indicators or "proxy variables", named at line 292. You mention 'approx spaces' or 'quantified conceptual overlaps', but it doesn't really enlighten the concept, if not explained in more detail. Possibly give an example from the land-use domain..
Response: This has been done in Lines 265-268.
255-256: orthogonal: do you mean weakly correlated? or qualitatively uncorrelated?
section 4.2:
Response: This has been clarified in the following way in lines 271-274 “Ideally, though not always possible, they should be unrelated and if possible orthogonal in terms of the characteristics (dimensions) they capture and explain, although recent work with data primitives has shown that orthogonality is less important in terms of discriminating power than first thought [62]” with reference [62] linking to Comber, A. and Kuhn, W., 2018. Fuzzy difference and data primitives: a transparent approach for supporting different definitions of forest in the context of REDD+. Geographica Helvetica, 73(2), pp.151-163.
line294: comment: could you use Table 1(continued) instead of Table 2?
Response: This has NOT been done. I am leaving this to the journal to manage their formatting – I could not find a satisfactory way to make the table run over 2 pages in Overleaf. Sadly 2 tables wah the most pragmatic option.
Tables1 and 2: interesting choices.
You should mention somewhere that not every data primitive (though quantified) is not of the same kind: example: additivity or comparability is ok with house price (if you mean average per square_foot) but not easy with income inequatity (Gini?); also: Professional occupation versus Low-skilled occupation versus Unemployment.
Response: this has been done – see below!
Also to mention: space and time resolution for each data primitive is important. Example: Population density and Population flux (update frequency is crucial).
Response: this has been done – see below!
- 4.3: Analysing State and Change
321-323: you say "at appropriate spatial and temporal resolutions". That's the point! You should develop a bit in relation with the different kind of data primitives, as mentioned in the previous remark.
Response: This has been done: The following has been included that addresses the last 3 comments in Section 4.3 and has been inserted as new paragraph in Line 339 “Inevitably these data are of different types and a number of questions remain at this stage. First, capturing data at appropriate spatial and temporal resolutions for each primitive is important with some primitives having greater critical update constraints than others, (Population density and Population flux, for example). Similarly, house price could be the average house price regardless of size (as is currently done in the UK), price per square metre or even price per bedroom. Others will be harder to partition. What for example are the professions that should be included in "Professional Occupations" or "Low Skilled Occupations" under Occupations? Should Income inequality be defined according to standard Gini coefficient measure or in a more relative manner? These are local application level decisions and any contributing data used to support or create a primitive can be retained for later changes in understanding or definition.”
333-337: by using "vectors of change" do you mean they are attached to a pixel representation (data cube) rather than a polygon representation?
Response: This has been clarified in lines 367-369 in the revised submission as follows: “In the change vector approach, the positions of each neighbourhood or area are determined in a multivariate feature space, and as new data becomes available, changes in position can be quantified using the change vector.”
- 4.4. Problems yet to be solved:
Ok: several above remarks could be addressed here, after having been mentioned earlier in your text. That section deserves some more details about the issues: possibly with reference to similar issues from land-cover ones.
Response: This has been done as described in other responses AND through the inclusion of a small illustrative case study after Section 4.3 and before the old Section 4.4
Review conclusion:
interesting paper, a critical review and a proposed perspective (vector of changes based on data primitives). However a few improvements are probably necessary before final approval for publication. What requires a "major revision".
Response: Thank you for your positive review and helpful suggestions.
Ref: you may consider adding in Section 2:
Douglas S. Massey and Nancy A. Denton. The Dimensions of Residential Segregation. Social Forces, Vol. 67, No. 2 (Dec., 1988), pp. 281-315.
The Dimensions of Residential Segregation (urbanpolicy.net)
Response: Thank you for this suggestion. This has been done but in the Conclusions section where the following sentence has now been included: “In many ways this approach operationalises the wider ideas behind the seminal work of Massey and Denton in 1988 in their exploration of the dimensions of segregation by taking advantage of our data rich era and extending into other area level processes”
Reviewer 2 Report
I doubt if the paper really fits the profile of the Geoinformation journal. I recognize the proposed method as a form of "qualitatification" of numerical data into the dictionary (feature-property) pair where the property is of qualitative form. A list of those pairs is presented in a form of a table, but that what is challenging is how to do such conversion. there is no case study, even preliminary, no scheme how to convert census data into such a tuple. I do not question the scientific quality of that paper (good review of the literature!), I doubt if the paper would be interesting directly for the geoinformation community. I think it would be redirected to a more demography-oriented journal. However, if the paper is to publish in IJGI it should be supplemented by at least a preliminary case study and comparison with existing approaches
Author Response
ijgi-1224200: Extending geodemographics using data primitives: a review and a methodological proposal
Reviewer 2
I doubt if the paper really fits the profile of the Geoinformation journal. I recognize the proposed method as a form of "qualitatification" of numerical data into the dictionary (feature-property) pair where the property is of qualitative form. A list of those pairs is presented in a form of a table, but that what is challenging is how to do such conversion. there is no case study, even preliminary, no scheme how to convert census data into such a tuple. I do not question the scientific quality of that paper (good review of the literature!), I doubt if the paper would be interesting directly for the geoinformation community. I think it would be redirected to a more demography-oriented journal. However, if the paper is to publish in IJGI it should be supplemented by at least a preliminary case study and comparison with existing approaches
Response: This has been done. A small illustrative case study has been included after Section 4.3 and before the old Section 4.4.
Reviewer 3 Report
The publication is interesting. The authors have presented data primitives as underlying dimensions or measurements that capture the characteristics of the process under investigation. They use a multidimensional feature space to quantify current state and changes in state. They can be used to create class classifications if required, as well as support predictive geodemography by modeling and analyzing state trajectories.
In their study, the authors proposed a set of primitives that could be used to characterize a range of social and economic processes that occur in neighborhoods.
However, the data in Tables 1 and 2 lack confirmation and would need to be supplemented with specific examples and then the publication would not only be a review of the literature and research, but also a confirmation of specific changes occurring in different environments and using data primitives for them.
The publication meets the requirements of the ISPRS International Journal of Geo-Information and can be accepted for publication with the suggested additions.
Author Response
ijgi-1224200: Extending geodemographics using data primitives: a review and a methodological proposal
Reviewer 3
The publication is interesting. The authors have presented data primitives as underlying dimensions or measurements that capture the characteristics of the process under investigation. They use a multidimensional feature space to quantify current state and changes in state. They can be used to create class classifications if required, as well as support predictive geodemography by modeling and analyzing state trajectories.
In their study, the authors proposed a set of primitives that could be used to characterize a range of social and economic processes that occur in neighborhoods.
However, the data in Tables 1 and 2 lack confirmation and would need to be supplemented with specific examples and then the publication would not only be a review of the literature and research, but also a confirmation of specific changes occurring in different environments and using data primitives for them.
The publication meets the requirements of the ISPRS International Journal of Geo-Information and can be accepted for publication with the suggested additions.
Response: we thank the reviewer for their positive comments. A small illustrative case study has been included after Section 4.3 and before the old Section 4.4.
Round 2
Reviewer 1 Report
Thank you for the improved manuscript, and satisfactory answers.
(typos: lines 267, 389?)
Reviewer 2 Report
The paper now is complete and definitely fits the geoinformation journal. I'm happy to recommend it for publication. With case study example entire method seems to be way much better
