Predictive Modeling of Customer Response to Marketing Campaigns

Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

Firstly, please accept my apologies concerning the computing the Accuracy in the previous review.

You are right, the values are 0.87 and 0.75 and not the ones that I mentioned in the first review.

 

Secondly, in this revised version, I recommend writing a  sentence about the information from

-  Table 1 (Line 194)

- Table 9 (Line 655)

Comments on the Quality of English Language

1) There are still words in British English even if most of them were replaced with their American correspondent word. For instance, “analyse” —Lines 113, 185, 597, and 637 (British English). The correct word in American English is “analyze”

 

2) Please revise the writing of:

- “NB and RF”. The correct form is “NB, and RF” (Line 137)

- “nearest”. The correct form is “the nearest” (Line 142)

- “86.8% ,”. The correct form is “86.8%,” (Line 162)

- “In this study”. The correct form is “In this study,” (Line 180 – the comma is missing)

- “of RF”. The correct form is “of the RF” (Line 181)

- “being”. The correct form is “is” (Line 209)

- “guarantee reproducibility”. The correct form is “guarantee the reproducibility” (Line 272)

- “model on four”. The correct form is “model in four” (Line 313)

- “Demographic factors such as age and income play significant roles”. The correct form is “Demographic factors, such as age and income, play significant roles” (Lines 577-578)

- “findings of the previous”. The correct form is “findings of previous” (Line 593)

- “in Appendix”. The correct form is “in the Appendix” (Line 600)

Author Response

Reviewer	Comments	Our Answer
Reviewer 1	Firstly, please accept my apologies concerning the computing the Accuracy in the previous review. You are right, the values are 0.87 and 0.75 and not the ones that I mentioned in the first review.   Secondly, in this revised version, I recommend writing a  sentence about the information from -  Table 1 (Line 194) - Table 9 (Line 655)	Thank you for your valuable suggestion. We have added a sentence in the text to provide further commentary on Table \ref{tab:tab:benchmark1} line 192, summarizing the key insights gained from the benchmark of related works. Thank you for your helpful suggestion regarding table 9. We have added a sentence summarizing the key insights from Table \ref{tab:benchmark} and pointed to Section 5.4.2 for a more detailed comparison and analysis.”
	2) Please revise the writing of: - “NB and RF”. The correct form is “NB, and RF” (Line 137) - “nearest”. The correct form is “the nearest” (Line 142) - “86.8% ,”. The correct form is “86.8%,” (Line 162) - “In this study”. The correct form is “In this study,” (Line 180 – the comma is missing) - “of RF”. The correct form is “of the RF” (Line 181) - “being”. The correct form is “is” (Line 209) - “guarantee reproducibility”. The correct form is “guarantee the reproducibility” (Line 272) - “model on four”. The correct form is “model in four” (Line 313) - “Demographic factors such as age and income play significant roles”. The correct form is “Demographic factors, such as age and income, play significant roles” (Lines 577-578) - “findings of the previous”. The correct form is “findings of previous” (Line 593) - “in Appendix”. The correct form is “in the Appendix” (Line 600)	Thank you for pointing out these corrections. We have carefully reviewed and revised all the suggested changes, including: Correcting 'NB and RF' to 'NB, and RF' (Line 137), Changing 'nearest' to 'the nearest' (Line 142), Fixing '86.8% ,' to '86.8%,' (Line 162), Adding the missing comma after 'In this study' (Line 180), Revising 'of RF' to 'of the RF' (Line 181), Correcting 'being' to 'is' (Line 209), Updating 'guarantee reproducibility' to 'guarantee the reproducibility' (Line 272), Adjusting 'model on four' to 'model in four' (Line 313), Adding commas in 'Demographic factors, such as age and income, play significant roles' (Lines 577-578), Revising 'findings of the previous' to 'findings of previous' (Line 593), Updating 'in Appendix' to 'in the Appendix' (Line 600). We appreciate your thorough review.

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

Please refer to the attachment.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Moderate editing of English language required.

Author Response

Reviewer 2	Firstly, the content of this article does not accurately address the main topic. Although the term 'marketing campaigns' appears in the body of the article, there is no clear discussion of its actual meaning or the activities involved. A more thorough exploration of what constitutes marketing campaigns, including examples and key strategies, would have enhanced the overall understanding and relevance of the article.	Thank you for pointing this out. We agree that a more thorough discussion of what constitutes marketing campaigns, along with examples and key strategies, would improve the clarity of the paper. We have added a more detailed explanation in the Introduction section to address this point. This enhancement will provide readers with a clearer understanding of the context in which our proposed solution operates.
	2. In regards to the related works, while the authors provide a table summarizing the information, the illustrations in this section are disorganized, making it difficult to connect them with the topics. If the sections were divided into several sub-titles, it might make the argument clearer.	We appreciate the reviewer's suggestion to improve the organization of the related works section. In response, we have restructured the section by dividing it into thematic subsections: \textit{Decision Tree Models}, \textit{Other Machine Learning Models}, and \textit{Hybrid Approaches and Ensemble Models}. These subsections provide a clearer organization of the reviewed studies and connect each work more directly to the focus of our research. We believe this restructuring improves the readability and coherence of the section. Additionally, we have retained Table \ref{tab:benchmark1} to summarize key findings for a quick reference.
	3. As shown in Figure 1, it occupies too much space and lacks necessary explanations. More effort is needed to provide comprehensive and valuable insights or viewpoints.	We appreciate your feedback regarding Figure 1. We understand that it occupies a considerable amount of space and lacks necessary explanations. In light of your comment, we aim to enhance the figure to provide more comprehensive insights. Additionally, we would like to mention that another reviewer recommended showing the interconnection between different components within the figure. We will take both comments into account to improve the clarity and effectiveness of Figure 1.
	4. When you first mention a term (e.g. DT, ML, SVM, NN, RF), please write out their entire names in full, and do not use any abbreviations. This is important to ensure that readers can easily understand and follow the text without any confusion.	We appreciate your valuable feedback regarding the use of abbreviations in our manuscript. We have addressed this issue by writing out the full names of all terms upon their first mention, ensuring that readers can easily understand and follow the text without any confusion. Thank you for your insight!
	5. Please delete any unnecessary tables, such as Table 4, as they are duplicated several times (e.g., Tables 7-8). Also, please enhance Figures 5-6 by including more information while simplifying the display of the four metrics (just showing the numbers of values). Meanwhile, please combine Figures 5-6 for comparison.	We appreciate your insightful feedback regarding Figures 5 and 6. In response, we have enhanced the figures by simplifying the display of the four metrics, focusing on the numerical values for clarity. Additionally, we have combined Figures 5 and 6 into a single figure for easier comparison of evaluation metrics before and after resampling. we also deleted table 4. The revised figure now provides a clearer visual representation of the metrics, facilitating better understanding of the changes. Thank you for your suggestion!
	6. As for 4.5 Decision Rules, the authors just wrote two lines. Please give us more illustrations.	We appreciate your feedback regarding Section 4.5 Decision Rules. We recognize that the current content may not provide sufficient detail. To enhance this section, we plan to include additional illustrations and examples of the decision rules derived from our model. This will include a more comprehensive discussion of how these rules operate within the context of our findings and their implications for practical marketing strategies. Thank you for highlighting this opportunity for improvement, and we look forward to providing a clearer and more informative section.
	7. Regarding section 5.3 Limitations and 5.4.2 Comparison with Our Proposed Solution, it appears that ChatGPT or Microsoft Copilot, or other AI tools, wrote the content. Regarding the type of paper writing, I disagree. Kindly rephrase the text using your own words, taking into account the conclusions of your research.	We appreciate your feedback regarding the limitations and comparison sections. we understand the importance of ensuring that our writing style meets the standards of academic rigor. To address your concerns, we have rephrased the content in these sections to enhance clarity and originality while maintaining the core insights of our study. Thank you for your valuable input, and we hope the revised text meets your expectations. To clarify, the limitations discussed in Section 5.3 have been articulated in general terms based on the findings of our research. Dataset Specificity: We acknowledge that our study utilizes a limited dataset specifically related to food companies, which constrains the generalizability of the model’s results to other sectors. Computational Efficiency Considerations: We emphasize the importance of detailing the computational efficiency of the models, especially in real-world scenarios. While we discuss the trade-off between computational demands and accuracy gains—particularly with Gradient Boosting compared to Decision Trees and Random Forests—we recognize the need for further elaboration on training times and resource utilization to provide practical insights. We have noted these limitations in the revised text and aimed to communicate them more effectively. Thank you for your valuable suggestion.

Author Response File: Author Response.pdf

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

The paper focuses on developing a predictive model using Decision Trees (DT) to analyze customer behaviour and improve marketing campaign targeting. The model aims to increase engagement and optimise ROI by incorporating customer demographics and purchase history data. The paper highlights the importance of factors like the recency of purchase, customer duration, and response history, providing actionable insights for marketing professionals.

Potential Flaws:

1. The appendix, particularly decision tree rules, could be enhanced by providing more structured or graphical representations to aid interpretability. For example, the decision rules could be better visualised with flowcharts instead of plain algorithms.

2. The dataset from the iFood platform is specific to a particular context and may limit the model's generalizability to other industries or regions. The authors should consider testing the model with datasets from diverse sectors for broader applicability.

3. The paper does not provide enough details on the model's computational efficiency, especially in real-world scenarios. Comparing training times and computational resources required could offer practical insights into its scalability.

Overal. This paper presents a well-structured approach to improving marketing campaign effectiveness through predictive modelling. Including resampling techniques and emphasis on feature importance add depth to the research. However, minor improvements, such as better graphical formatting in the appendix and a more detailed discussion of the model's generalizability, would enhance its clarity and practical applicability. I recommend this paper for publication with minor revisions.

Author Response

Reviewer 3	The paper focuses on developing a predictive model using Decision Trees (DT) to analyze customer behaviour and improve marketing campaign targeting. The model aims to increase engagement and optimise ROI by incorporating customer demographics and purchase history data. The paper highlights the importance of factors like the recency of purchase, customer duration, and response history, providing actionable insights for marketing professionals. Potential Flaws: Overal. This paper presents a well-structured approach to improving marketing campaign effectiveness through predictive modelling. Including resampling techniques and emphasis on feature importance add depth to the research. However, minor improvements, such as better graphical formatting in the appendix and a more detailed discussion of the model's generalizability, would enhance its clarity and practical applicability. I recommend this paper for publication with minor revisions.
	1. The appendix, particularly decision tree rules, could be enhanced by providing more structured or graphical representations to aid interpretability. For example, the decision rules could be better visualised with flowcharts instead of plain algorithms.	Thank you for your suggestion. We have updated the first algorithm in the appendix to be presented as a flowchart for enhanced interpretability. We believe that having both graphical and algorithmic representations can offer a more comprehensive understanding of the decision tree rules. Therefore, while we have changed the first algorithm to a flowchart, we opted to retain the remaining algorithms in their original format. If preferred, we are open to updating the rest of the algorithms as flowcharts as well.
	2. The dataset from the iFood platform is specific to a particular context and may limit the model's generalizability to other industries or regions. The authors should consider testing the model with datasets from diverse sectors for broader applicability.	You are completely correct, and we appreciate your insightful comment. We fully recognize the limitation of using a dataset specific to the iFood platform, and we agree that testing the model with datasets from diverse sectors would enhance its generalizability. We have already mentioned in the conclusion section that we plan to extend this research by testing the model with datasets from various industries and regions to ensure broader applicability.
	3. The paper does not provide enough details on the model's computational efficiency, especially in real-world scenarios. Comparing training times and computational resources required could offer practical insights into its scalability.	Thank you for your insightful comment. We acknowledge the importance of detailing the computational efficiency of the models, especially in real-world scenarios. As noted in the paper, we mention the trade-off between computational demands and accuracy gains, particularly with Gradient Boosting compared to Decision Trees and Random Forests. However, we agree that further elaboration on training times and resource utilization would offer more practical insights. We plan to extend our work by conducting more comprehensive tests on different datasets to evaluate computational performance. And we added sentence to the conclusion clearifiying this idea as future plan

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

The author has made efforts to provide relevant explanations, but these additions do not significantly enhance the overall quality of the paper. Additionally, the excessive use of yellow highlighting makes the text difficult to read. I recommend that the authors revise the paper, addressing these issues, and resubmit it for further review.

Comments on the Quality of English Language

 Extensive editing of English language required.

Author Response

Reviewer	Comments	Our Answer
Reviewer 1	Improve the research desgin	In response to your feedback regarding the need to improve the research design, I have refined the section to provide a clearer and more systematic approach to how the study addresses the identified challenges and evaluates the effectiveness of the Decision Tree (DT) model. Class Imbalance Addressed: I emphasized the critical issue of class imbalance identified in the literature. The revised design now clearly outlines how resampling techniques were applied to enhance the model's predictive performance. Evaluation of Effectiveness: The design includes a detailed evaluation of the DT model using key performance metrics—accuracy, precision, recall, and F1-score—before and after resampling. This illustrates the significant improvements achieved and reinforces the effectiveness of the model. Feature Importance Analysis: I incorporated a focus on feature importance analysis to identify the key factors influencing customer responses to marketing campaigns. This section now highlights the significance of demographic factors and past interactions, providing insights into customer behavior and effective marketing strategies. Overall, these enhancements aim to clarify the research design and better demonstrate how the study's findings contribute to understanding customer responses to marketing campaigns
	methods adequately described	Thank you for your constructive feedback regarding the descriptions of the methods in the Data Collection and Model Selection subsections. To address your comments and improve clarity, I have made several enhancements: Data Collection Subsection: I provided a more detailed account of the data collection process, including the specific sources of the data, the criteria for selecting the dataset, and the rationale behind these choices. Additionally, I included information on the data preprocessing steps that were applied prior to model training, ensuring transparency in the data preparation process. Model Selection Subsection: I expanded on the explanation of the Decision Tree (DT) model, elaborating on its structural characteristics and the reasons for its selection over other models such as Support Vector Machines (SVM) and Neural Networks (NN). This includes a discussion on the balance between interpretability and performance, highlighting how the DT model’s intuitive "if-else" logic makes it particularly suitable for marketing applications, where stakeholders may have limited technical knowledge. Furthermore, I clarified the trade-offs involved in model selection, emphasizing the importance of understanding the motivations behind predictions in marketing contexts. These revisions aim to provide a more comprehensive understanding of the methods employed in the study and their relevance to the research objectives. I appreciate your guidance in enhancing the rigor of this work.
	The conclusion supported by the results comment	We appreciate your insightful comment regarding the need for a stronger linkage between the conclusions and the results presented in our study. In response, we have revised the discussion section to ensure that each conclusion is explicitly tied to the corresponding results. We have added clarifying sentences to highlight how the findings support our conclusions, providing a more coherent narrative that enhances the overall clarity and impact of our work. Thank you for your valuable feedback, which has helped us improve the rigor of our research.

We sincerely appreciate your thoughtful feedback and recommendations for improving our paper. While your comment did not specify particular sections or subsections to revise, we took your concerns to heart and have made comprehensive efforts to enhance the clarity and coherence of the entire manuscript. We have revised various subsections to ensure that our explanations are more relevant and impactful.

Furthermore, we acknowledge your point regarding the excessive use of yellow highlighting, which may hinder readability. The highlighting was used solely to indicate changes made in response to reviewer comments. However, we have addressed this issue by reducing the highlighting and refining the overall presentation of the text to improve the reader's experience.

Thank you again for your valuable insights, which have guided us in enhancing the quality of our paper. We look forward to your feedback on our revised submission.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

Please refer to the attachment.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Minor editing of English language required.

Author Response

Reviewer 1	In section 5.4.2 , the lack of explanation accompanying table 8 necessitates further elaboration to contextualise the findings presented. please corect the edition	We thank the reviewer for their helpful suggestion. We have now expanded the explanation accompanying Table 8 in Section 5.4.2 to provide better context and a more thorough discussion of the findings. Specifically, we have elaborated on how the accuracy of the models compares to related work, the effectiveness of handling imbalanced datasets, and the trade-offs between accuracy, complexity, and interpretability. These additional details clarify the insights presented in the table and align them with the key themes of our study.
		\noindent We thank the reviewer for their observation. The content previously outlined in the "Comparison with Our Proposed Solution" section was intended to compare our findings with the existing literature presented in Table 8. Based on the reviewer's comment, we have paraphrased this section to better align the discussion with the thematic focus of the paper. We believe this revision clarifies the relevance of these points and their connection to our work.
		We appreciate your valuable feedback. To clarify, the first table in the paper serves as a benchmark of the existing literature on predictive models for direct marketing, offering an overview of key studies and their findings. In contrast, Table 8 is designed to compare our proposed solution with those benchmarked models. Additionally, we have updated the caption of Table 8 to more accurately reflect this comparison.

Author Response File: Author Response.pdf

Round 4

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

The authors have made a lot of effort to revise their manuscript according to the reviewer’s comments 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

very interesting topic of the article, but its extension is questionable.

first of all, it would be good to mention in the introduction or methodology that the study is based on work that has already been published - your editor should also have this.

Abstract

It is certainly necessary to add the necessary goal and its connection to the achieved results.

In the abstract, you point to specific data, but you should also link it and explain your article's methodology, or the questions and their answers.

Introduction

In the introduction, you should explain more about the problem, the problematic situation. Subsequently, it is necessary to solve and formulate research questions. I propose that the introductory part of the text be divided into: introduction, theoretical solution and methodology. This way it will be clear to the reader what the article will be about and who you researched the given problem.

The results

In section 1, you present three research questions – how are these question connected with results sectionl? (more specific and clarify...)

Discussion

It would certainly be appropriate to add limitations.

Conclusion

Answering questions cannot be just pointing to section in the text, it is necessary to answer questions specifically and relevantly, or to support it or not.

Resources

It was appropriate to expand the resources in the article with similar research and investigations from the point of view of current authors (last 3 years) and their results in this area.

Other points:

Table 1 and 9 are identical, why do you present the same data.

Extension of the text - if this contribution is to be an extension of the original, it should contain more data or more new parts in each chapter of the text, not only in the discussion. If I understand it correctly, the authors only completed the part in the discussion to expand the article.

best regards, 

Comments for author File: Comments.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper focuses on predicting key factors that influence customer response to marketing campaigns.

The topic is relevant to the field. Firstly, the customer response to marketing campaigns can be measured with different tools, and the Decision Tree (DT) chosen by the authors is a suitable approach. Secondly, customer response to any marketing campaign is not easy to assess, and the results can be objective when different tools are employed.

The paper implements the Decision Tree model for predicting customer response to marketing campaigns and it is assessed by the confusion matrix.

The methodology is designed properly. All the stages of the methodology are detailed and there is a coherent information flow among them.

The conclusions must be revised to be consistent and correlated with the results obtained because the computed errors that I highlighted impact on results and conclusions.

The references are appropriate according to the topic of the research.

The authors must:
- cite in text Table 1, e.g., “Table 1 presents …..”
-  comment on the information from Table 1 in one or more sentences
- center “Hardware” and “Software” in the cell, in Table 2
- mention the source of Table 3
- draw arrows among text boxes and inside them, in Figure 1, to underline the flow
- revise the values from Figures 2 and 3 because there are errors in calculating them

 

1) 2. Related Work

- I think that is better the title in the pl;ural form: “Related Works”

- Please split the paragraphs into two or more paragraphs so it can be easier to read. Follow the rule “keep one idea to one paragraph”. Otherwise is very difficult to follow the ideas by readers.

2) 3. Proposed Solution

- Please draw arrows among text boxes and inside them, in Figure 1, to underline the flow

- Please cite in text Table 1, e.g., “Table 1 presents …..”

- Please comment on the information from Table 1 in one or more sentences

3) 3.1. Hardware and Software Configuration

- Please center “Hardware” and “Software” in the cell, in Table 2

4) 3.2. Data Collection

 - Please use “,” separator for 1,000 units for the 2206 value (Line 153)

- You should mention once again the source of Table 3

5) 3.4. Data Preprocessing

- Please split the paragraphs into two or more paragraphs so it can be easier to read.

6) 3.5.1. Hyperparameter-tuning

- Please split the paragraphs into two or more paragraphs so it can be easier to read (Lines 213-241)

7) 4.2.2. After Resampling

- You mention that the model became more balanced (Line 366). However, the model is not so balanced taking into account that FP=24, and this value is half of TP=49

8) 4.3.1. Before Resampling

- Please revise the values from Figure 2 because there are errors in calculating them:

* Accuracy: (27+21) / (27 + 21 + 36 + 357) = 48/ 441= 0.108

* Precision: 0.56 (is correct)

* Recall: 0.428

* F1-Score: 0.486

- Please revise the comments concerning these values, as well

9) 4.3.2. After Resampling

- Please revise the values from Figure 3 because there are errors in calculating them:

* Accuracy: 0.544

* Precision: 0.67 (is correct)

* Recall: 0.830

* F1-Score: 0.742 (is correct)

- Please revise the comments concerning these values, as well

10) 5.1. Results Interpretation

- Please revise the comments taking into account the corrected values

- Please split the paragraphs into two or more paragraphs so it can be easier to read (Lines 421-449)

11) 5.2. Implications for Marketing Strategies

- Please split the paragraphs into two or more paragraphs so it can be easier to read (Lines 451-487)

12) 5.3.1. Overview of Related Works

- Please replace “table below” with “Table 9”

Comments on the Quality of English Language

1) The paper is written in both British and American English language!

It is critical to choose what type of English (either American English or British English) you want to write your paper and use it on the entire paper before submitting the revised paper.

For instance, there are:

- “behaviour” —Lines 3, 15, 27, etc. (British English). If you decide to use American English, the correct word is “behavior”

- “behaviours” —Line 486, etc. (British English). If you decide to use American English, the correct word is “behaviors”

- “analyze” —Lines 105, 131, and 472. (American English). If you decide to use British English, the correct word is “analyse”

- “modelling” — Title of the paper, Lines 19, 52, etc. (British English). If you decide to use American English, the correct word is “modeling”

 

3) Revise the English language syntax, for instance in Line 144 – “In Table 2 are listed the specific components and tools used”. In this case, the order is noun and verb, so the correct sentence is “In Table 2 the specific components and tools used are listed”

 

2) Please revise the writing of:

- “improves”. The correct form is “to improve” (Line 16)

- “the 36 false negatives (FN) produced, indicate cases where ”. The correct form is “the 36 false negatives (FN) indicate cases where” (Line 147)

- Remove the colon “:” from the title of the subsection “4.2.2. After Resampling:” (Line 354)

- “responses improves”. The correct form is “responses improve” (Line 358)

- Remove the colon “:” from the title of the subsection “4.3.2. After Resampling:” (Line 388)

- “influential to customer”. The correct form is “influential on customer” (Line 408)

- “findings of previous”. The correct form is “findings of the previous” (Line 467)

- “approach in enhancing”. The correct form is “approach to enhancing” (Line 538)

Reviewer 3 Report

Comments and Suggestions for Authors

This paper presents a contribution of using predictive models for direct marketing. The paper is well written and well-structured. The authors show transparently their hardware and software set-up, as well the data collection mechanism. They justify the selection of the  decision tree (DT) supervised Machine Learning (ML) method. The authors also explain their model development approach by explaining their data partitioning approach : 80% for training  and 20% for evaluation. The research also takes account the reproducibility of the research.  The authors  resampled their data for improving true positives with also a drop in the number of false negatives. Also, they improved their metrics regarding precision and recall. That is aligned with the second confusion matrix.  The authors also provide their decision rules transparently.   In particular, the authors show that their Decision Tree solution with Gradient boosting outperformed previous approaches. Specifically in imbalance datasets, which are common in direct marketing sources.  The authors also empathise that the interpretability of decision trees is often preferable. 

The authors need to fix minor presentation issues such as:

Line 48: A space needed between "DT" and the citation.

Line 66 and 67: Too much space between the questions.

Figure 1: The lines of the figure have no direction. It is not clear if it is a sequential procedure (from the picture).

Reviewer 4 Report

Comments and Suggestions for Authors

 

Market predicting modelling is clearly an area of important practical applications that is attracting a lot of academic and practitioner’s interest. Therefore, the thematic of the paper is rather current and interesting. The authors clearly defined the three questions that they are trying to explain. It is good that the authors also clearly presented their code in the appendix.

 

I have some comments that I think could help improve the paper:

 

1)       For a paper of this type, 22 references is a bit low with some important related papers not mentioned.

2)       I think that an important part is understanding better the data, which is key for this type of analysis. The authors use the Kagel database, which is well known, but I think that should add a more detailed description of the data, with perhaps some histograms, tables or other suitable metrics. It is difficult to interpret the results without having a more details understanding of the data.

3)       I think that it is also important to justify why the authors use Decision Trees rather than any of the many other available techniques. To make the interpretation easier? To get more accurate results? DT are a reasonable choice but need to be supported (explained) a bit more.

4)       A brief style suggestion. Some of the section could be divided into paragraph to make reading a bit easier. For example, the Related Work section is interesting but difficult to read because it is one (massive) paragraph.

5)       There should be some more details about the limitation of the paper. For example, some comments about the sample size?

 

Minor comments:

1)       Check the abbreviations list I think that there are some abbreviations not used in the paper.

2)       The presentation of some of the tables could be improved

Comments on the Quality of English Language

Overall the English is adequate.