Non-Destructive Viability Discrimination for Individual Scutellaria baicalensis Seeds Based on High-Throughput Phenotyping and Machine Learning

Round 1

Reviewer 1 Report

1.       Authors are suggested to improve the introduction section by adding some more literature related to machine learning and hyperspectral imaging-based seed viability analysis.

2.       Materials and methods section should include the information on regional viability seeds, collection approach and its storage during experiments.

3.       Figure 1 legends is not self-explanatory.

4.       Among all the spectral pre-processing techniques why SNV, DT, and MSC are selected? Also, authors have used SMOTE method. What are the other available methods for data balancing? Clear justification for selecting any particular approach should be there.

5.       What hyperparameters were considered in case of Random Forest (e.g., number of estimators, number of leaf nodes etc.). Also mention the grid values for them.

6.       The interval of parameter division is too coarse (i.e., 0.01, 0.1, 1, and 10), which may not guarantee that the SVM model can obtain the best value. What is the effect of increasing the grid range?

7.       English is modest. Spacing, punctuation marks, grammar, and spelling errors should be reviewed thoroughly. Many typos and incomplete sentences are found throughout the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This article is about Non-destructive viability discrimination for individual Scutellaria baicalensis seeds based on high-throughput phenotyping and machine learning. I have several questions:

1. What are the Values of different parameters in SVM?

2. What are the Values of different parameters in RF?

3. Please explain the ratio train, validation, and test sets from all data.

 

4. Please also use different criteria for classification.

Author Response

Please see the attachment.

Author Response File: Author Response.docx