Next Article in Journal
Is It Possible to Forecast the Price of Bitcoin?
Next Article in Special Issue
Attention-Based CNN-RNN Arabic Text Recognition from Natural Scene Images
Previous Article in Journal
A Time-Varying Gerber Statistic: Application of a Novel Correlation Metric to Commodity Price Co-Movements
Previous Article in Special Issue
Queue Length Forecasting in Complex Manufacturing Job Shops
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forecasting
Volume 3
Issue 2
10.3390/forecast3020023
Review Report
forecasting-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
Article
Peer-Review Record

A New Machine Learning Forecasting Algorithm Based on Bivariate Copula Functions

Forecasting 2021, 3(2), 355-376; https://0-doi-org.brum.beds.ac.uk/10.3390/forecast3020023
by J. A. Carrillo 1, M. Nieto 1, J. F. Velez 2 and D. Velez 1,*
Reviewer 1: Anonymous
Reviewer 2:
M. Ishaq Bhatti
Forecasting 2021, 3(2), 355-376; https://0-doi-org.brum.beds.ac.uk/10.3390/forecast3020023
Submission received: 26 April 2021 / Revised: 21 May 2021 / Accepted: 24 May 2021 / Published: 27 May 2021
(This article belongs to the Special Issue Forecasting with Machine Learning Techniques)

Round 1

Reviewer 1 Report

This is an interesting topic. I have several questions on this paper which need clarification prior to getting the "green light" for acceptance. Please note and address them below.

(1) The risk references are incomplete. Please add some more modern references to make the paper more accurate such as the following:

Guharay, Sabyasachi, K. C. Chang, and Jie Xu. "Robust Estimation of Value-at-Risk through Distribution-Free and Parametric Approaches Using the Joint Severity and Frequency Model: Applications in Financial, Actuarial, and Natural Calamities Domains." Risks 5, no. 3 (2017): 41.

(2) Line 188: "Selection of the copula function C which best models the
relationship between explanatory variables and the residual term obtained in the previous step." What are the range of copula functions which are allowed? There are many types including empirical copulas. So what are the eligible ones?

(3) From Figure 1: Why is the Expected value used solely? For data with high outliers, why isn't Median being used?

(4) Equation (12) notation for d_epsilon is not clear.

(5) For Table 1, why is AIC being chosen instead of BIC?

(6) Figure 3 is based on exactly what? A simulation example? Real life data? Further clarity is needed.

(7) In Algorithm 2, what is the precise definition of the auxError?

(8) For the Results section: Are there are any simulated datasets where the results are known? This is crucial in the Verification & Validation steps. I only see the Validation steps using the UCI Machine Learning data. Where is this? If not used, how can you scientifically justify skipping this?

 

Please answer the above eight questions.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Please see the attached report

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I am satisfied with the responses to my questions. I still feel that better verification & validation (V&V) could be done, but for the time being it is acceptable.

Back to TopTop