FP-Growth Algorithm for Discovering Region-Based Association Rule in the IoT Environment

Round 1

Reviewer 1 Report

Dear Authors,

I enjoyed reading your paper.

Still, there are some presentation issues to deal with. For instance:

• The paper must follow the specific structure of the journal:

https://www.mdpi.com/journal/mathematics/instructions namely:
Introduction, Materials and Methods, Results, Discussion, Conclusions, etc.;

• English language and style issues - Grammarly on default settings detected only for the Title + Abstract part 4 correctness alerts and some other issues related to engagement, word choice, passive voice misuse, unclear sentences, comma misuse between sentences (a score of 77 out of 100). Moreover, since you do not appear to be native English speakers, I recommend Extensive editing of English language and style (at least the Grammarly full report for the entire paper with a maximum score or the “No issues found” message);
• Other language-related issues (e.g. at the end of the label for Figure 2. An example of transaction database and ordered); 
• The lack of explanations for certain abbreviations (e.g. K - on the top right of Table 2) means that the reader should guess what it is about (I suggest a footnote briefly specifying “thousand”).

Thank you very much for your contribution.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposes an RFP-Growth algorithm which is a sort of improved FP Growth based on spatial-temporal association data mining algorithm. The paper presents detailed descriptions of the research background, problem definition and algorithm design. And the experiment results also verify that the RFP-Growth algorithm can solve the problem that the original FP growth algorithm cannot find new association rules in the whole data containing location information. However, this study still has some shortcomings. It is recommended that the authors make some essential amendments to the following points to meet the acceptance and publication criteria.

1. The full name of the FP abbreviation should be written as “Frequent Pattern” when it first appears in the abstract and the main text of the paper.

2. The contents of Figure 2, Figure 3 and Figure 4 are the same, except that the figure captions are different. This is very confusing and unusual. And the main text of the paper does not describe the sub-graphs in Figure 3 and Figure 4. Please revise this issue.

3. In section 3.1 overview of algorithm, the paper assumes that the transaction raw data analyzed contain longitude (x) and latitude (y). The Region-based FP-Growth algorithm designed in the paper will use the DBSCAN algorithm to cluster the transaction area. However, the x and y in transactions with position data in Figure 5 are not like latitude and longitude data.

4. In paragraph 3.2, the Improved FP-Growth algorithm designed by this paper is based on the intersection of grouping transaction items. However, the algorithm still lacks some descriptions of how the algorithm handles when there is no intersection between transaction items or when the content of the intersection is empty. Especially, when the amount of transaction data is sparse, there will often be no intersection between transaction items. How will the Region-based FP-Growth algorithm proposed in the paper solve this problem? Please add an explanation.

5. On section 4.1.2 data set, the experiment of this paper is based on real transaction data and synthetic data. Although the authors described the form of the two types of data. However, authors should provide snippet of samples of the two types of data including descriptions of the data fields.

6. The experimental performance analysis of the RFP-Growth Tree algorithm still lacks the comparison result of the performance analysis of the memory usage. The authors are also recommended to provide more comprehensive experimental performance analysis results, such as minsup vs. memory usage in terms of Frequent Itemset Generation/Closed Frequent Itemset Generation/Maximal Frequent Itemset Generation/Rare Itemset Generation. The authors should supplement the experimental analysis results in this regard.

7. In addition, there are many related studies on FP-Growth based on spatial-temporal association data mining algorithm. It is recommended that the authors provide some performance comparison results among designed RFP-Growth algorithm and state-of-the-art algorithms to verify the performance superiority of the RFP-Growth algorithm.

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper proposes a RFP growth algorithm to support data mining. The authors claims, it is noble. However, there exists number of concerns which should be addressed to improve the quality of the paper. Figures 2, ,3 and 4 appear same. It is difficult for us to understand. There are some grammatical errors found which should be addressed. Although in the introduction authors mentioned IoT and the availability of huge amounts of spatial data, this necessitates the development of new algorithms to support data mining techniques.  However, the authors are unable to present robust cases by taking account of real world problems how the proposed algorithm work. Therefore, to validate the proposed algorithm a real world case study like the data of certain city where various shipping malls exist can be considered. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors clarified and responded to the questions in my review report one by one, and the revised manuscript has been significantly improved. I am satisfied with the result of the correction. Therefore, this revised paper can be accepted for publication.

Reviewer 3 Report

The authors address all the concerns.