TAID-LCA: Segmentation Algorithm Based on Ternary Trees
Round 1
Reviewer 1 Report
Introduction: the last 2 paragraphes need to be written to strengthen the motivation and clarify what's new to be proposed
In troduction: last sentence: what does TAID stand for
Sec 4: can sec's 4.1 and 4.3 be clarified to be linked or one based on the other?
Sec 4.2: Authors names need to be properly written using say Olmus instead fo OLMUS
Sec's 4.2 and 4.3: clarify any differences or connections between the two sec's
Sec 6: why sec 6.1 is used, with no sec 6.2
Sec 7: can the first two paragraphs on R be simplified/clarified, in terms of what's new to be proposed in this manuscript
Sec 7: what's the message of the last sentence on the software still in test phase; does this make any sense to implement TAID-LCA
Example(s): can any examples be included
Sec 8: is a concluding sec needed to summarize the findings or comments from the manuscript
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
1. It is suggested to add comparative analysis of the proposed method with existing methods. 2. Implementation on the real data can be discussed stepwise in the current paper. 3. For the practitioner, I can suggest providing all R codes in the appendix if it is not possible, then an open access Github repository can be provided for implementation. 4. I did not find the summary and conclusion section in the article. 5. What are the limitations and future recommendations of the stated study?Author Response
The paper has been updated and an application with real data was included. Additionally, we added conclusions and stated that this proposal is only for the use of categorized variables and will work in the future for continuous variables.
Reviewer 3 Report
After carefully reading the proposed paper, the results are technically sound, and the paper is well written and organized. I recommend the publication of this paper after some minor comments:
My comments are:
- Expand the abstract Section to be more fitted.
- An abbreviation Section should be added.
- The outline of the paper should be included.
- More information around figure 1 should be reported.
- Are the algorithms herein can be used in cases of categorical data, especially, nominal kind?
- What about the partition of the chi-square technique? Please, explain.
- How we can apply the ratio test based on these algorithms?
- Conclusion remakes should be reported based on important results.
- The application section should be analyzed to prove the theoretical sections.
- What about the logistic regression based on these algorithms? Explain a simple example.
- The references are very few and they contain many old references, many modern references must be mentioned and more details should be mentioned in the introduction.
- Check for typographical errors.
Author Response
The paper has been updated and an application with real data was included. Additionally, we added conclusions and stated that this proposal is only for the use of categorized variables and will work in the future for continuous variables.
- More information around figure 1 should be reported.
Complementary information was added
- Are the algorithms herein can be used in cases of categorical data, especially, nominal kind?
Only in the explanatory or predictive variables.
- What about the partition of the chi-square technique? Please, explain.
It is implicitly explained in the introduction referencing to Kass (1980) where the use of chi-square only detects symmetric relationships between variables.
- How we can apply the ratio test based on these algorithms?
It can be an area to explore for future research projects.
- What about the logistic regression based on these algorithms? Explain a simple example.
We can use it to assess or compare our segmentation proposal. For example, to assess suitably classified individuals.
Reviewer 4 Report
I recomend the paper for a minor revision with the following comments considered.
Comments
1) An example of a data sample and the result of the algorithm should be included in the paper.
2) An additional section "Conclusions" should be added to the paper, which includes a summary of the results and possibile directions for a future work.
3) The paper needs a revision. Some sentences may be shortened. Suggestions for revision are given bellow (underlined).
- (page 2) detect interactions, they state that this is due to the
- (page 2) Recently Gunduz and Lutfi (2021) present important contribution QUEST algorithm that decreases uncertainty to compare with CHAID.
- (page 3) The latent class models help to statistical segmentation process which the response variables have heterogeneity in the groups (Gon¸calves et al., 2020)
- (page 4) table have not a symmetrical role, i.e. one is conditioned
- (page 5) to replace the chi-square test, as we propose
- (page 7) These conditions can be modified according to the particularities of the analysis to be performed.
- (page 7) a model tree too fitted to data, it may be excessively complex
- (page 7) this process is generally called post-pruning, inspired in the act of cutting of branches from a tree.
- (page 8) Use all the data for training and apply an statistical test
- (page 10) whose elimination produces the greater improve in precision (but only if the elimination of some one produces an improvement, otherwise the rule remains intact)
- (page 10) The name and inspiration of the SA method arise
- (page 11) that use some library like poLCA
Comments for author File: Comments.pdf
Author Response
The paper has been updated and an application with real data was included. Additionally, we added conclusions and stated that this proposal is only for the use of categorized variables and will work in the future for continuous variables. There has been a proper revision of English as suggested. We have identified precise areas of work in the future and appreciate your suggestions.
Round 2
Reviewer 2 Report
I would like to congratulate the authors for their excellent work. They have made a very detailed revision and convinced me with good arguments. Therefore, I recommend this article for the possible publication in Mathematics.