Scholarlink

<p>For agricultural products, the critical drop height (CDH) refers to the minimum height a&#xa0;product can fall without serious damage. This is especially important for sensitive fruits like dragon fruit, which are prone to impact-related spoilage. Determining CDH enables better handling practices from harvest to consumption, preserving quality and minimizing losses. Instead of costly and time-consuming lab tests, CDH can be estimated using fast and affordable methods. This study aims to predict the CDH of dragon fruit using ridge regression (RR) and artificial neural networks (ANNs), evaluated by mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R<sup>2</sup>) metrics. For test partition, while the results of the RR model’s MAE: 13.94, RMSE: 17.70, and R<sup>2</sup>: 0.78, the ANN model’s results are MAE: 9.33, RMSE: 13.42, and R<sup>2</sup>: 0.87. For both models, plots were drawn using the predicted values and actual values of the models. With the results of the RR model, the effects of the predictor variables on CDH were shown, elasticities were calculated, and the regression equation was written. RR model results indicate, while compression force, toughness, mass, and hardness are significant, compression deformation, energy, brix, major diagonal diameter, minor diagonal diameter, sphericity, area, volume, and perimeter are insignificant when predicting CDH of dragon fruit.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ridge Regression and Artificial Neural Network Model Comparison for Predicting Critical Drop Height in Dragon Fruit Based on Selected Fruit Characteristics

  • 期刊论文

  • 发表日期：

    2025年6月24日

查看全文
  • Uğur Ercan,
  • Onder Kabas,
  • Aylin Kabaş

摘要

For agricultural products, the critical drop height (CDH) refers to the minimum height a product can fall without serious damage. This is especially important for sensitive fruits like dragon fruit, which are prone to impact-related spoilage. Determining CDH enables better handling practices from harvest to consumption, preserving quality and minimizing losses. Instead of costly and time-consuming lab tests, CDH can be estimated using fast and affordable methods. This study aims to predict the CDH of dragon fruit using ridge regression (RR) and artificial neural networks (ANNs), evaluated by mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R2) metrics. For test partition, while the results of the RR model’s MAE: 13.94, RMSE: 17.70, and R2: 0.78, the ANN model’s results are MAE: 9.33, RMSE: 13.42, and R2: 0.87. For both models, plots were drawn using the predicted values and actual values of the models. With the results of the RR model, the effects of the predictor variables on CDH were shown, elasticities were calculated, and the regression equation was written. RR model results indicate, while compression force, toughness, mass, and hardness are significant, compression deformation, energy, brix, major diagonal diameter, minor diagonal diameter, sphericity, area, volume, and perimeter are insignificant when predicting CDH of dragon fruit.