LSTM-Based Deep Learning Models for Long-Term Tourism Demand Forecasting

Round 1

Reviewer 1 Report

The authors implemented two forecasting models for long-term tourism demand based on LSTM networks. They evaluated the two models at first on real hotel booking data obtained from three hotels in Greece, and then on these hotel data combined with weather data obtained from the online weather data repository Weather Underground.

I enjoyed reading this paper. It is very well written, the methodology is properly described as the results.

I have only a few comments to improve the readability of the paper:

1. In the Introduction, some references should be added in the first two paragraphs when you refer to the proposed models for tourism demand forecasting.
2. Table 1 is not cited in the text.
3. In Section 3, you refer to Hotel A, B and C in the text and in three tables without having explaining what they are. This explanation comes in fact in Section 4 (Data). The same thing applies to weather data. You should consider to move Section 4 before Section 3.
4. The acronyms in the items of Subsection 5.2 (Benchmark models) can be removed from the title and left only in the text.
5. To make a better comparison among the results, Figures 7 and 8 can become two subfigures of one figure. The same thing can be applied to (a) Figures 9, 10, 11, (b) Figures 3, 4, 5. Moreover, colors in Figures 7 and 8 should be changed because the differences among the models cannot be grasped.
6. It would be interesting as future work to compare these results with those of hotels in other areas such as Italy.

  

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Its better to revise the abstract by including challenges, proposed solution, and shortcomings. The text in the Figure seems like image. Its better to improve the quality of figures. Its better to improve the quality of flow chart in the Figure 2. Authors mention that they used the data for different seasons for hotels. Its better to include comparison season wise. The comparison with different forecasting models can be improved by including details from, Optimal BRA based electric demand prediction strategy considering instance‐based learning of the forecast factors. It’s suggested to refine abstract, title, and conclusion by adding more details. How authors claim their approach is beneficial for researchers?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I have no further comments.