Design of a Waste Classification System Using a Low Experimental Cost Capacitive Sensor and Machine Learning Algorithms
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsRemark 1:
To prepare better paper will be good to provide more information in state-of-the-art about other solution using AI/ML. There are mentioned such solutions but more information about: method which they used, the advantages of using it, the results which they provide using this method could be interesting for the readers.
Remark 2:
The authors could judge why they choose Random Forest algorithms, more explanation about could be interesting.
The answer for question: • What is the main question addressed by the research? The main question address the sensor construction with AI/ML method for waste detection. • Do you consider the topic original or relevant to the field? Does it address a specific gap in the field? Please also explain why this is/ is not the case. The topic is valuable according to waste recycle process used right now, so new, cheap method for waste detecting/recogizing are necessary to develop. The authors propose their own construction and used well know ML method. • What does it add to the subject area compared with other published material? The authors proposed their own construction, which is interesting according to possible implementation. • What specific improvements should the authors consider regarding the methodology? What further controls should be considered? The methodology used is ok, their use appropriate amount of data and different objects. • Are the conclusions consistent with the evidence and arguments presented and do they address the main question posed? Please also explain why this is/is not the case. The conclusions are consistent with the evidence and arguments presented in the paper. • Are the references appropriate? The references are appropriate, but I have add a remarks that authors could provide more information about different ML method used for waste recognition. • Any additional comments on the tables and figures. The data and results are presented in good way.Author Response
Attached is a document that responds to the observations made by the evaluator.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsDear Authors,
To better evaluate the manuscript, I have separated it into topics that highlight the merit of the article for publication in Applied Science. Additionally, I suggest reviewing and adjusting 5 items that I believe need to be reviewed to make the article more accessible to the different readers of MDPI - Applied Science.
Favorable Points for Publication in MDPI's Applied Sciences
Innovative Contribution: The manuscript proposes a low-cost capacitive sensor for waste classification, which addresses an important global challenge—solid waste management. The approach combines recycling principles and machine learning, aligning well with Applied Sciences’ focus on practical, innovative solutions;
Sustainability Focus: The use of recycled aluminum and low-cost materials aligns with current trends in environmental sustainability and supports the journal's interest in advancing practical, eco-friendly technologies.
Strong Results and Validation: The experimental results demonstrate a robust performance of the proposed sensors, especially the MNT model, which shows high sensitivity and precision in classifying waste. The integration with machine learning further validates the approach.
Broad Applicability: The proposed system has potential applications in smart cities and industrial waste management, extending the relevance of the research beyond academia to practical implementations.
Clear Alignment with Scope: The interdisciplinary approach, blending material science, sensor design, and artificial intelligence, fits well within the scope of MDPI’s Applied Sciences.
Main Points to Adjust for Publication in MDPI's Applied Sciences
Clarity in Figures and Diagrams: Some diagrams, such as Figures 2 and 3 (sensor designs), lack sufficient detail or annotations to make them easily interpretable for readers unfamiliar with capacitive sensors. Including clearer labels, descriptions, and visual enhancements would improve comprehension.
Expand the Discussion Section: While the results are strong, the discussion could better contextualize the findings in relation to existing technologies. Comparing the performance and costs of the proposed sensors with commercial alternatives or other research in the field would strengthen the manuscript.
Improve Methodological Transparency: The methods section could benefit from additional details about the experimental setup and machine learning model. For example: a) Specify how the sample size (n = 100) was chosen; b) Detail the parameters used in the Random Forest algorithm; c) Explain the rationale for selecting capacitance as the primary feature.
Address Potential Limitations: The manuscript does not adequately address limitations or challenges, such as: a) Sensitivity to environmental factors (e.g., humidity, temperature) that might affect sensor performance; b) Potential scalability issues or long-term durability of the sensors. Including this analysis would make the paper more balanced and credible.
Language and Formatting Adjustments: The text contains occasional grammatical errors and instances of awkward phrasing (e.g., "describes a better behavior"). These should be revised to ensure professional and fluent language throughout. Additionally, ensure adherence to MDPI's Applied Sciences formatting guidelines, particularly regarding references, figures, and captions.
If these adjustments are addressed, the manuscript will have a higher chance of being accepted for publication in Applied Sciences.
Best Regards
Reviewer
Author Response
Attached is a document that responds to the observations made by the evaluator.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe research presents the design of low-cost capacitive sensor systems for solid waste classification. The study introduces two sensor prototypes: the Traditional Model (MT) and the Non-Traditional Model (MNT). These sensors are built using recyclable materials and are designed to identify types of wastes as plastic, glass, metal, and organic waste. The sensors utilize changes in capacitance to detect the material type. In the process of validating MNT being more sensitive and dynamic range for application, the statistical procedures and the adopted Machine Learning method are questionable.
1. The stability of MNT and MT have not been tested as with replicated reading of capacitance for the same waste sample in different shapes under varied conditions.
2. Part 3.3, Why is a smaller capacitance reading an indicator of better performance in this hypothesis? Besides, a t-test is used to determine whether two different distributions are significantly different.
3. The capacitance reading for organic waste is notably different from the readings of other types (10^2 vs 10^1), there seems to be no need to adopt the random forest algorithm for classification using capacitance reading values as direct input.
Author Response
Attached is a document that responds to the observations made by the evaluator.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsThe statistical methodology applied here was not reasonble to support the aming conclusion. The use of RF for classification make no sense as the reading value itself is significant enough to tell the conclusion.
Author Response
The observations made were addressed in the best way and are described in the attached document.
Author Response File: Author Response.pdf
