Single and Multiple Separate LSTM Neural Networks for Multiple Output Feature Purchase Prediction

Round 1

Reviewer 1 Report

1- The introduction: please add some research question(s) and the main contributions of the study.

2- Related works: most of cited studies are published in the years 2016 to 2019 , more critics/discussions/investigation of pros and cons should be added to the investigation of these researches.

3- Section 4 (Dataset and data preprocessing): why dedicating a separate section to present the data while section 6 (results) details the tests and findings of this data.

4- Despite section 4 is dedicated to present the used dataset, it is not clear of the used data is public from the state-of-the-art (in this case, you should give the reference/link). If the used data is yours: is the data available online ?

5- For better clarity, can you transform some or all tables 1- 6 to histograms.

6- In the related works, authors discuss several lstm algorithms for purchase prediction problem.   Is there any contribution in the used methodology (in terms of the used LSTM)?: what is the novelty of the present work regarding the discussed studies?

7- Present the used lstm architecture.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper is concerned with using Long Short-Term Memory Neural Networks (LSTM NN) to purchase prediction. The paper is well-structured, including the presentation of related works, LSTM NN, methodology, results, and conclusion. However, there are some issues that should be improved. In my opinion, the main limitation of this research is used only one tool (LSTM NN) to purchase prediction. The results should clearly indicate that the use of LSTM NN is profitable (i.e., the accuracy of prediction for LSTM NN is higher than for other prediction models based, for example, on regression analysis or other types of neural networks). Moreover, the Conclusion section should present implications and limitations of using the proposed approach, for example, from the theoretical and practical point of view. I also suggest adding to the abstract main findings obtained in this study. Moreover, the abbreviation “LSTM” is used the first time in the abstract, so it needs to be explained there. The paper includes some language errors, so it should be also carefully proofread.  

For example:

- "various researches" in the first sentence of Introduction

- "were present" in page 11

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Present tense and passive voice should be used.

The abstract should be extended with the main findings, novelties and practical implications.

Key words may not duplicate the words in the title. It is a good idea to extend the focus of the paper.

The introduction section starts with literature review on the topics: prediction, forecasting.

The title of section "2. Related work" may be changed to "2. Literature review". The text in this section discusses different types of neural networks.

Section 3 is "3. LSTM neural networks". It describes the essence of LSTM NN.

Below figures and tables should be given the source - authors' contribution or cited (if it is allowed) or modified (if it is allowed).

Uniqueness of text: 95%.

Section 4 is "Dataset and data processing". The authors may give new figures giving better presentation of the results.

Since the literature review is in several sections, I recommend the use of the IMRAD format for structuring the paper. The introduction section should focus on the current state and the problem. The literature review should be another section.

The text with well-known info about the essence of LSTM NN should be changed with focused text on the application of the LSTM NN in the current research of the authors.

The methodology section follows the data section. It is not a good approach. Using the IMRAD format, the article will be OK.

Sentences with "our", "we" should be changed to Passive Voice.

Numeric example with the prediction should be given.

 

 

 

Present tense and passive voice should be used.

Sentences with "our", "we" should be changed to Passive Voice.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I can recommend the manuscript for publishing in the present form.

Author Response

Thank you very much for your review and positive decision.