Modify Leave-One-Out Cross Validation by Moving Validation Samples around Random Normal Distributions: Move-One-Away Cross Validation

Round 1

Reviewer 1 Report

The essence of manuscript is the synthesis of a new method of approximation of a complex mathematical function by moving a validation point around random normal distributions rather than leaving it out, named move-one-away cross validation, named MOA-CV method.
1. The content of the manuscript is very dispersed in its considerations and unnecessarily extended to 36 pages of text. In its present form, the manuscript is a comprehensive research report rather than a synthetic approach to the essence of scientific achievement.
2. I propose in section 4.2, Table 4, reducing the number of 20 functions tested to the most representative two LD and two HD.
3. I suggest resigning from appendixes A1 and A2, and put the formulas of the four chosen functions directly in the manuscript text.
4. Include only substantive conclusions resulting from the conducted research, giving up trivial conclusions, for example - "From the results of selecting best models, it is conclude that accurate rates of best model selection of MOA-CV and LOO-CV rise with the increase in samples sizes from 5n to 10n almost for any function, and MOA-CV performs slightly better than LOO-CV in selecting best models ", and "MOA-CV and LOO-CV methods performs much better in LD functions than in HD functions".

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper describes a novel modification of the leave-one-out cross validation.

The paper is very interesting; however, the paper is not well written. In particular, there are a lot of long sentences which are very hard to understand or sentences that should be rephrased.

Concerning the propose method, in Section 3, authors declared that the parameter lambda is set by hand. I think that it should be discuss how to set this parameter. Did the authors try different values? This parameter is equal for all the case studies considered for the validation?

At the end of Section 1 (the fourth line of page 3) Authors wrote about “two engineering problems” but in the other Sections only one problem is considered.

Concerning the Appendix A.2, this table is very hard to read for non-experts in the considered engineering problem. Is it really necessary to include it?

 

Minor points and typos

- In the abstract: “We modifies” -> “We modify”, “is unreliable of LOO-CV” -> “is unreliable in LOO-CV”.

- Page 2, line 1 “Monto Carlo” -> “Monte Carlo”.

- In Section 4 there is an ”Error! Reference source not found!” message.

- Page 15, line 3, “the LOO-CVerror indicate the PRS is” -> “the LOO-CVerror indicate that the PRS is”.

- Some references are not cited in the main text (this can be due to the error in section 4).

- Please, do not use the capital letter after the colon.

- Please do not use the contract forms of the verbs.

- Please, use the correct style for references (e.g. “Stone et al. [1]” not “Stone 1”).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors took all my comments into account and perfectly prepared the manuscript for publication.