See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/379309563

Classification of Apple Quality Using XGBoost Machine Learning Model

XGBoost Makine Öğrenimi Modelini Kullanarak Apple Kalitesinin
Sınıflandırılması

Conference Paper · March 2024

CITATIONS READS

0 42

1 author:

Muhammet Çakmak
Sinop Üniversitesi
24 PUBLICATIONS 83 CITATIONS

SEE PROFILE

All content following this page was uploaded by Muhammet Çakmak on 27 March 2024.

The user has requested enhancement of the downloaded file.

 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

4th International Conference on Innovative Academic Studies

Classification of Apple Quality Using XGBoost Machine Learning Model

Muhammet Çakmak*1
1Sinop University, Department of Computer Engineering, Sinop, [email protected]

Keywords Abstract
Classification It is important for the agricultural sector to determine the quality of apples by using the
Machine Learning properties of the external and internal structure of the apple together. Size, weight,
Apple Quality juiciness, sweetness, ripeness, crispness and acidity are used to determine the quality of
the apple and whether it is edible or not. These characteristics can also be used to
determine whether the apple can be used for juice production. Both the external
appearance and other internal qualities of apples directly affect apple quality. In this
study, we classify size, weight, juiciness, sweetness, ripeness, crispness and acidity using
machine learning methods to determine apple quality. In this study, XGB, Support Vector
Machine, KNN and RF methods are used to classify apple quality. XGB achieved the best
cross validation value with 87.44%, while RF achieved the best ROC AUC value with
77.94%.

XGBoost Makine Öğrenimi Modelini Kullanarak Apple Kalitesinin Sınıflandırılması

Anahtar Kelimeler Öz
Sınıflandırma Elmanın dış ve iç yapısının özelliklerinin birlikte kullanılarak elmanın kalitesinin
Makine öğrenme belirlenmesi tarım sektörü için önemlidir. Boyut, ağırlık, sululuk, tatlılık, olgunluk,
Elma Kalitesi çıtırlık ve asit değeri elmanın kalitesine ve yenilebilir olup olmadığana karar verilmesi
için kullanılır. Ayrıca bu özellikler elmanın meyve suyu yapımında kullanılıp
kullanılamayacağı içinde kullanılabilir. Elmaya ait hem dış görünüş hem de elmanın içine
ait diğer nitelikler elma kalitesini doğrudan etkilemektedir. Bu çalışmada elma kalitesini
belirlemek için boyut, ağırlık, sululuk, tatlılık, olgunluk, çıtırlık ve asit değerlerini makine
öğrenmesi yöntelerini kullanarak sınıflandırıyoruz. Çalışmada XGB, Support Vector
Machine, KNN ve RF yöntemleri kullanılarak elma kalitesi sınıflandırılmıştır. XGB
%87.44 ile en iyi cross validation değerine ulaştırken RF ise %77.94 ile en iyi ROC AUC
değerin elde etmiştir.

1
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

Introduction

Apples, esteemed as one of the most widely cultivated and consumed fruits globally, hold a profound
significance not only in the agricultural sector but also in human health and nutrition (Hu et al. 2024). Their
ubiquity in temperate climates and year-round availability in markets worldwide underscore their economic
importance and cultural relevance (El-Mesery et al. 2023).
Nutritionally, apples are lauded for their rich dietary fiber, essential minerals, and diverse array of vitamins,
making them a wholesome addition to any diet (Chen et al. 2024; Wójcik, Filipczak, and Wójcik 2024). Their
consumption is associated with a myriad of health benefits, ranging from bolstering antioxidant defenses to
fortifying immune function and promoting longevity (Ren et al. 2021; Wang et al. 2022). Scientific studies have
elucidated the role of apples in mitigating the risk of various diseases, including cardiovascular ailments and
certain types of cancer (Roy et al. 2022; Sair et al. 2023). Notably, polyphenols found abundantly in apples, such
as hydroxycinnamic acids, anthocyanins, flavan-3-ols/procyanidins, dihydrochalcones, and flavonols, are credited
with conferring these protective properties (Lopez-Rodulfo et al. 2024; Pu et al. 2022).
The global apple industry is anchored by China, which accounts for more than half of the world's apple
production (Zhao, Li, and Hu 2023). This dominance underscores the fruit's economic significance and its status
as a dietary staple for millions. Moreover, the versatility of apples, which can be consumed fresh, juiced, dried, or
processed into various products, further enhances their appeal and marketability (Gao et al. 2024; Ru et al. 2023).
While the external appearance of apples is often the first point of attraction for consumers, it is the internal
quality and texture that ultimately influence taste preferences and consumption patterns (Lagerkvist et al. 2023;
Tireki 2021). Parameters such as size, weight, juiciness, sweetness, ripeness, crispness, and acidity are crucial
indicators of apple quality, guiding purchasing decisions and consumer satisfaction (Wang et al. 2021; Xia et al.
2021; Zhang et al. 2021).
Rapidly developing technology has caused artificial intelligence applications to affect all sectors rapidly
(Dhanush et al. 2023; Schramm, Wehner, and Schmid 2023). To date, machine learning and deep learning
applications have been used in many fields, such as technology (Çakmak and Albayrak 2022; Dhanush et al. 2023),
health(Lassau et al. 2020), economy (Sharma et al. 2022), commerce (Wang et al. 2023), cyber security (Altunay
and Albayrak 2023), education (Lim 2024) and agriculture (Oyebisi and Alomayri 2023). Determining the quality
of apples through the observation and experience of an experienced expert is a process that takes a long time and
lacks accurate prediction (Çetin et al. 2022). In this study, apple quality will be determined using XGB, SVM, KNN
and RF machine learning methods. Thus, results were obtained at less cost, in a shorter time and with high
accuracy.

Material and Method

Datasets

The Apple Quality dataset (Anon 2024) offers a comprehensive exploration of key characteristics defining the
quality of apples. These attributes encompass size, weight, sweetness, crunchiness, juiciness, ripeness, acidity, and
overall quality. Size serves as a visible indicator influencing consumer preference, while weight provides insights
into density and juiciness. Sweetness, a fundamental taste trait, dictates flavor profiles, while crunchiness and
juiciness contribute to texture and sensory appeal. Ripeness, denoting maturity, impacts flavor development, and
shelf life. Acidity balances sweetness, enhancing flavor complexity. Overall quality synthesizes these factors,
reflecting the holistic desirability of apples. The dataset can be accessed from Kaggle and contains 4000 pieces of
data. Figure 1 shows the data distribution of the dataset.

Figure 1. Data distribution of the apple quality dataset

2
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

Figure 2 shows sample data of the dataset.

Figure 2. Sample data of the apple quality dataset

Machine Learning Methods

Extreme Gradient Boosting (Chen and Guestrin 2016)(XGB) is a powerful machine learning algorithm
renowned for its exceptional performance in classification and regression tasks. Based on decision trees, XGB
sequentially builds a strong ensemble model by iteratively correcting the errors of preceding trees. Through
gradient boosting, XGB optimizes the loss function, emphasizing the importance of misclassified instances. This
iterative approach enables XGB to effectively capture complex patterns and interactions within the data, making
it particularly adept at handling large datasets with high-dimensional features. Moreover, XGB incorporates
regularization techniques to mitigate overfitting, ensuring robust generalization to unseen data. Its versatility,
efficiency, and scalability have established XGB as a popular choice in various domains, from finance to healthcare
and beyond.
Support Vector Machine (SVM) is a widely used supervised learning algorithm recognized for its effectiveness
in both classification and regression tasks. SVM operates by identifying the optimal hyperplane that maximizes the
margin between different classes in the feature space. By transforming the input data into a higher-dimensional
space through kernel functions, SVM achieves nonlinear decision boundaries, enabling it to handle complex
relationships and achieve high classification accuracy. SVM is particularly advantageous in scenarios with small to
medium-sized datasets, where it excels at separating classes even in the presence of noisy or overlapping data
points. Its robustness, versatility, and ability to handle both linear and nonlinear problems have cemented SVM as
a cornerstone of machine learning algorithms.
K-Nearest Neighbors (KNN) is a simple yet powerful algorithm for classification and regression tasks,
particularly in scenarios where the underlying data distribution is not well-defined or exhibits nonlinear
relationships. KNN operates by assigning a class label to a given data point based on the majority vote of its nearest
neighbors, with "k" representing the number of neighbors considered. By measuring the similarity between data
points using distance metrics such as Euclidean or Manhattan distance, KNN determines the nearest neighbors
and predicts the target variable accordingly. Despite its simplicity, KNN can effectively capture intricate patterns
and adapt to diverse datasets. However, its computational complexity grows with the size of the dataset, making
it less suitable for large-scale applications.
Random Forest (RF) is an ensemble learning algorithm that leverages the power of decision trees to achieve
robust and accurate predictions. RF constructs multiple decision trees independently using bootstrapped samples
of the training data and random subsets of features. Through ensemble averaging, RF aggregates the predictions
of individual trees to produce a final outcome, thereby reducing overfitting and improving generalization
performance. Additionally, RF introduces randomness during tree construction, enhancing diversity among trees
and promoting exploration of different feature combinations. This inherent diversity, coupled with the ability to
handle both classification and regression tasks, renders RF highly versatile and resilient to noisy or correlated
input features. Its scalability, interpretability, and ability to capture complex interactions make RF a popular choice
across various domains, from bioinformatics to marketing and beyond.
In this study, precision, accuracy, f1-score and recall performance values of XGB, SVM, KNN and RF machine
learning algorithms for Apple Quality were compared.

Experimental Study

Performance Metrics

A multi-class evaluation was conducted on the apple quality dataset, which contained 7 categories. The models'
performance was evaluated using the True Positive (TP), False Positive (FP), True Negative (TN), and False
Negative (FN) values obtained from the confusion matrix. Equations (1), (2), (3), (4), (5), (6), and (7) show the
values obtained. F1-Score (F1Sco), Accuracy (Acc), Sensitivity (Sen), and Precision (Pre) metrics were used to
compare the models' performance. These metrics were calculated using equations (1), (2), (3), (4), (5), (6), and
(7).
3
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

For a class k,
(1)

(2)

(3)

(4)

(5)

(6)

(7)

Results and Discussions

In this study, we aimed to determine the quality of apples by leveraging both external and internal properties
such as size, weight, juiciness, sweetness, ripeness, crispness, and acidity. These characteristics are crucial for
assessing the edibility of apples as well as their suitability for juice production. By employing machine learning
techniques including XGBoost (XGB), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Random
Forest (RF), we sought to classify apple quality effectively. Table 1 shows confusion matrix, cross validation and
ROC AUC curve values of XGBoost. Table 1 shows Xgboost's confusion matrix and cross-validation values, while
Figure 3 shows XGBoost's ROC AUC curve values.

Table 1. Confusion matrix and cross validation of XGBoost.

precision recall f1-score support

0 0.80 0.75 0.77 608

1 0.76 0.80 0.78 592

accuracy 0.78 1200

macro avg 0.78 0.78 0.78 1200
weighted avg 0.78 0.78 0.78 1200

Cross Validation Score : 87.44%

Figure 3. ROC AUC values of XGBoost

4
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

The XGBoost model exhibits strong performance in classifying apple quality, as evidenced by the classification
report and evaluation metrics. With precision scores of 0.80 and 0.76 for classes 0 and 1 respectively, it accurately
identifies instances of each class. The recall scores of 0.75 and 0.80 indicate its ability to capture a high proportion
of actual instances for both classes. The F1-scores around 0.78 suggest a balanced performance in terms of
precision and recall. The overall accuracy of 78% demonstrates the model's capability to classify instances across
the dataset correctly. Additionally, a cross-validation score of 87.44% highlights its consistency in performance
across different data subsets, indicating robustness and generalization. Moreover, the ROC_AUC score of 77.78%
underscores its effectiveness in discriminating between positive and negative classes. In summary, the XGBoost
model offers a reliable classification of apple quality, providing valuable insights for stakeholders in the
agricultural sector.
Table 2 shows SVC's confusion matrix and cross-validation values, while Figure 4 shows SVC's ROC AUC curve
values.

Table 2. Confusion matrix and cross validation of SVC.

precision recall f1-score support

0 0.72 0.70 0.71 608

1 0.70 0.73 0.71 592

accuracy 0.71 1200

macro avg 0.71 0.71 0.71 1200
weighted avg 0.71 0.71 0.71 1200

Cross Validation Score : 78.93%

Figure 4. ROC AUC values of SVC

With precision scores of 0.72 and 0.70 for classes 0 and 1 respectively, it accurately predicts instances of each
class, while the recall scores of 0.70 and 0.73 suggest its ability to capture a substantial proportion of actual
instances for both classes. The F1-scores around 0.71 reflect a balanced performance in terms of precision and
recall. The overall accuracy of 71% indicates the model's capability to correctly classify instances across the
dataset. Additionally, a cross-validation score of 78.93% highlights its consistency in performance across different
data subsets, indicating reasonable robustness and generalization. However, the ROC_AUC score of 71.10%
suggests a moderate ability to discriminate between positive and negative classes.
Table 3 shows KNN's confusion matrix and cross-validation values, while Figure 5 shows KNN's ROC AUC curve
values.

5
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

Table 3. Confusion matrix and cross validation of KNN.

precision recall f1-score support

0 0.78 0.74 0.76 608

1 0.75 0.79 0.77 592

accuracy 0.76 1200

macro avg 0.76 0.76 0.76 1200
weighted avg 0.77 0.76 0.76 1200

Cross Validation Score : 85.62%

Figure 5. ROC AUC values of KNN

The KNN model demonstrates precision scores of 0.78 and 0.75 for classes 0 and 1 respectively, it accurately
predicts instances of each class. The recall scores of 0.74 and 0.79 indicate its ability to capture a significant
proportion of actual instances for both classes. The F1-scores around 0.76 reflect a balanced performance in terms
of precision and recall. The overall accuracy of 76% suggests the model's capability to classify instances across the
dataset correctly. Furthermore, a cross-validation score of 85.62% underscores its consistency in performance
across different data subsets, indicating robustness and generalization.
Table 4 shows RF's confusion matrix and cross-validation values, while Figure 6 shows RF's ROC AUC curve
values.

6
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

Table 4. Confusion matrix and cross validation of RF.

precision recall f1-score support

0 0.84 0.70 0.76 608

1 0.73 0.86 0.79 592

accuracy 0.78 1200

macro avg 0.79 0.78 0.78 1200
weighted avg 0.79 0.78 0.78 1200

Cross Validation Score : 85.45%

Figure 6. ROC AUC values of RF

RF model demonstrates precision scores of 0.84 and 0.73 for classes 0 and 1 respectively, it accurately predicts
instances of each class. The recall scores of 0.70 and 0.86 suggest its ability to capture a significant proportion of
actual instances for both classes. The F1-scores around 0.76 and 0.79 reflect a balanced performance in terms of
precision and recall. The overall accuracy of 78% underscores the model's capability to classify instances across
the dataset correctly. Moreover, a cross-validation score of 85.45% emphasizes its consistency in performance
across different data subsets, indicating robustness and generalization. Notably, the ROC_AUC score of 77.94%
signifies a strong ability to discriminate between positive and negative classes.

Discussion

In this comprehensive comparative analysis evaluating the efficacy of machine learning models for the
classification of apple quality, XGBoost and RF emerge as the most robust performers, showcasing superior
performance across multiple evaluation metrics when compared to SVC and KNN models.
XGBoost demonstrates competitive precision and recall scores, with class-specific precision rates of 80% and
76%, and recall rates of 75% and 80%, respectively. These metrics indicate its ability to classify instances from
both classes accurately. The F1-scores around 78% reflect a harmonious balance between precision and recall,
while an overall accuracy of 78% underscores its proficiency in correctly classifying instances. Additionally,
XGBoost's impressive cross-validation score of 87.44% highlights its consistency and generalization ability across
diverse data subsets.
Similarly, RF exhibits strong precision and recall rates, with class-specific precision scores of 84% and 73%,
and recall scores of 70% and 86%, respectively. The balanced F1-scores around 76% and 79% further underscore
its effectiveness in classification tasks. With an overall accuracy of 78% and an exceptional ROC_AUC score of
77.94%, RF demonstrates robust discriminative ability, making it a compelling choice for apple quality
classification.
Conversely, SVC and KNN models display moderate performance, with lower precision, recall, and ROC_AUC
scores compared to XGBoost and RF. SVC achieves precision scores of 72% and 70%, with recall scores of 70%
and 73%, while KNN exhibits precision scores of 78% and 75%, with recall scores of 74% and 79%. Though both
models demonstrate reasonable accuracy, their discriminative ability, as indicated by the ROC_AUC scores (SVC:
71.10%, KNN: 76.45%), falls short of the performance achieved by XGBoost and RF.

7
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

In summary, the comprehensive evaluation underscores the superior performance of XGBoost and RF models
in apple quality classification tasks, offering stakeholders in the agricultural sector valuable insights for optimizing
processes and enhancing apple quality assessment methodologies.

Conclusion

Upon thorough evaluation, our comprehensive analysis concludes that XGBoost and RF stand out as the most
proficient models for apple quality classification, exhibiting superior performance when compared to SVC and
KNN models across multiple evaluation metrics. XGBoost demonstrates commendable precision and recall rates,
indicating its ability to effectively classify instances from both classes, while maintaining a balanced F1-score
around 78%. With an overall accuracy of 78% and a remarkable cross-validation score of 87.44%, XGBoost proves
its consistency and generalization capability across diverse data subsets. Similarly, RF showcases strong precision
and recall rates, complemented by an exceptional ROC_AUC score of 77.94%, highlighting its robust discriminative
ability. These models offer valuable insights for stakeholders in the agricultural sector, enabling them to optimize
processes and enhance apple quality assessment methodologies. In contrast, SVC and KNN models, while
demonstrating moderate performance, fall short in terms of discriminative ability and overall effectiveness when
compared to XGBoost and RF. Thus, our analysis suggests that XGBoost and RF models are the preferred choices
for apple quality classification tasks, providing reliable and accurate predictions that can significantly benefit
agricultural practices and decision-making processes.

References
1. Altunay, Hakan Can, and Zafer Albayrak. 2023. “A Hybrid CNN + LSTMbased Intrusion Detection System for Industrial IoT Networks.”
Engineering Science and Technology, an International Journal 38. doi: 10.1016/j.jestch.2022.101322.Anon. 2024.
“Https://Www.Kaggle.Com/Datasets/Nelgiriyewithana/Apple-Quality.”
2. Çakmak, Muhammet, and Zafer Albayrak. 2022. “AFCC-r: Adaptive Feedback Congestion Control Algorithm to Avoid Queue Overflow in
LTE Networks.” Mobile Networks and Applications 27(5). doi: 10.1007/s11036-022-02011-8.
3. Çetin, Necati, Kevser Karaman, Erhan Kavuncuoğlu, Bekir Yıldırım, and Ahmad Jahanbakhshi. 2022. “Using Hyperspectral Imaging
Technology and Machine Learning Algorithms for Assessing Internal Quality Parameters of Apple Fruits.” Chemometrics and Intelligent
Laboratory Systems 230:104650. doi: 10.1016/J.CHEMOLAB.2022.104650.
4. Chen, Shuaihong, Shaowu Zhang, Tiantian Hu, Hui Li, Jianxi Sun, Guangzhao Sun, and Jie Liu. 2024. “Responses of Soil Reactive Nitrogen
Pools and Enzyme Activities to Water and Nitrogen Levels and Their Relationship with Apple Yield and Quality under Drip Fertigation.”
Scientia Horticulturae 324:112632. doi: 10.1016/J.SCIENTA.2023.112632.
5. Chen, Tianqi, and Carlos Guestrin. 2016. “XGBoost: A Scalable Tree Boosting System.” Proceedings of the ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining 13-17-August-2016:785–94. doi: 10.1145/2939672.2939785.
6. Dhanush, Guduru, Narendra Khatri, Sandeep Kumar, and Praveen Kumar Shukla. 2023. “A Comprehensive Review of Machine Vision
Systems and Artificial Intelligence Algorithms for the Detection and Harvesting of Agricultural Produce.” Scientific African 21:e01798. doi:
10.1016/J.SCIAF.2023.E01798.
7. El-Mesery, Hany S., Kwami Ashiagbor, Zicheng Hu, and W. G. Alshaer. 2023. “A Novel Infrared Drying Technique for Processing of Apple
Slices: Drying Characteristics and Quality Attributes.” Case Studies in Thermal Engineering 52:103676. doi: 10.1016/J.CSITE.2023.103676.
8. Gao, Qingchao, Yingxin Wang, Yahui Li, Weikang Yang, Wayne Jiang, Ying Liang, and Zhiyong Zhang. 2024. “Residue Behaviors of Six
Pesticides during Apple Juice Production and Storage.” Food Research International 177:113894. doi: 10.1016/J.FOODRES.2023.113894.
9. Hu, Yaohua, Yichen Qiao, Bingru Hou, Zhanghao Qu, Peng Zhang, Runzhe Han, and Jiapan Guo. 2024. “Building Models to Evaluate Internal
Comprehensive Quality of Apples and Predict Storage Time.” Infrared Physics & Technology 136:105043. doi:
10.1016/J.INFRARED.2023.105043.
10. Lagerkvist, C. J., A. K. Edenbrandt, L. A. Bolos, and R. M. Nayga. 2023. “Consumer Acceptance of Aesthetically Imperfect Vegetables – The
Role of Information Framing and Personal Values: Evidence from the United States.” Food Quality and Preference 104:104737. doi:
10.1016/J.FOODQUAL.2022.104737.
11. Lassau, N., I. Bousaid, E. Chouzenoux, J. P. Lamarque, B. Charmettant, M. Azoulay, F. Cotton, A. Khalil, O. Lucidarme, F. Pigneur, Y. Benaceur,
A. Sadate, M. Lederlin, F. Laurent, G. Chassagnon, O. Ernst, G. Ferreti, Y. Diascorn, P. Y. Brillet, M. Creze, L. Cassagnes, C. Caramella, A.
Loubet, A. Dallongeville, N. Abassebay, M. Ohana, N. Banaste, M. Cadi, J. Behr, L. Boussel, L. Fournier, M. Zins, J. P. Beregi, A. Luciani, A.
Cotten, and J. F. Meder. 2020. “Three Artificial Intelligence Data Challenges Based on CT and MRI.” Diagnostic and Interventional Imaging
101(12):783–88. doi: 10.1016/J.DIII.2020.03.006.
12. Lim, Eun Mee. 2024. “Metaphor Analysis on Pre-Service Early Childhood Teachers’ Conception of AI (Artificial Intelligence) Education for
Young Children.” Thinking Skills and Creativity 51:101455. doi: 10.1016/J.TSC.2023.101455.
13. Lopez-Rodulfo, Ivan M., Emmanouil D. Tsochatzis, Emil W. Stentoft, Pamela Martinez-Carrasco, Julia D. Bechtner, and Mario M. Martinez.
2024. “Partitioning and in Vitro Bioaccessibility of Apple Polyphenols during Mechanical and Physiological Extraction: A Hierarchical
Clustering Analysis with LC-ESI-QTOF-MS/MS.” Food Chemistry 441:138320. doi: 10.1016/J.FOODCHEM.2023.138320.
14. Oyebisi, Solomon, and Thamer Alomayri. 2023. “Potential Application of Artificial Intelligence to the Alpha and Gamma Radiation from
Agricultural Byproducts Used as Building and Construction Materials.” Scientific African 21:e01792. doi: 10.1016/J.SCIAF.2023.E01792.
15. Pu, Yijing, Xu He, Luyao Chen, Hongxuan Wang, Yuxia Ma, and Weibo Jiang. 2022. “Apple Polyphenols Attenuate the Binding Ability of
Angiotensin Converting Enzyme 2 to Viral Proteins: Computer Simulation and in Vitro Experiments.” Food Bioscience 50:102090. doi:
10.1016/J.FBIO.2022.102090.
16. Ren, Guangxin, Xueling Ran, Ruiyuan Zeng, Jianwei Chen, Yanbo Wang, Chunlan Mao, Xiaojiao Wang, Yongzhong Feng, and Gaihe Yang.
2021. “Effects of Sodium Selenite Spray on Apple Production, Quality, and Sucrose Metabolism-Related Enzyme Activity.” Food Chemistry
339:127883. doi: 10.1016/J.FOODCHEM.2020.127883.
17. Roy, Souvik, Anil Kumar Mondru, Tania Chakraborty, Abhijit Das, and Sandipan Dasgupta. 2022. “Apple Polyphenol Phloretin Complexed
with Ruthenium Is Capable of Reprogramming the Breast Cancer Microenvironment through Modulation of PI3K/Akt/MTOR/VEGF
Pathways.” Toxicology and Applied Pharmacology 434:115822. doi: 10.1016/J.TAAP.2021.115822.
18. Ru, Xiaoya, Jie Zhou, Kaiyuan Gong, Zhihao He, Zhanwu Dai, Meirong Li, Xinxin Feng, Qiang Yu, Hao Feng, and Jianqiang He. 2023. “Climate
Warming May Accelerate Apple Phenology but Lead to Divergent Dynamics in Late-Spring Frost and Poor Pollination Risks in Main Apple
Production Regions of China.” European Journal of Agronomy 150:126945. doi: 10.1016/J.EJA.2023.126945.
8
 4th International Conference on Innovative Academic Studies (ICIAS) – 12-13 March 2024 – Konya, Turkey

19. Sair, Ali Tahir, Yitong Li, Weiyang Zhao, Tong Li, and Rui Hai Liu. 2023. “Anticancer Activity of Apple Peel Extracts against Human Breast
Cancer Cells through Insulin-like Growth Factor-1 Signal Transduction Pathway.” Journal of Agriculture and Food Research 11:100507.
doi: 10.1016/J.JAFR.2023.100507.
20. Schramm, Simon, Christoph Wehner, and Ute Schmid. 2023. “Comprehensible Artificial Intelligence on Knowledge Graphs: A Survey.”
Journal of Web Semantics 79:100806. doi: 10.1016/J.WEBSEM.2023.100806.
21. Sharma, Manu, Sunil Luthra, Sudhanshu Joshi, and Anil Kumar. 2022. “Implementing Challenges of Artificial Intelligence: Evidence from
Public Manufacturing Sector of an Emerging Economy.” Government Information Quarterly 39(4):101624. doi:
10.1016/J.GIQ.2021.101624.
22. Tireki, Suzan. 2021. “A Review on Packed Non-Alcoholic Beverages: Ingredients, Production, Trends and Future Opportunities for
Functional Product Development.” Trends in Food Science & Technology 112:442–54. doi: 10.1016/J.TIFS.2021.03.058.
23. Wang, Chenxing, Sayed Fayaz Ahmad, Ahmad Y. A. Bani Ahmad Ayassrah, Emad Mahrous Awwad, Muhammad Irshad, Yasser A. Ali, Muna
Al-Razgan, Yasser Khan, and Heesup Han. 2023. “An Empirical Evaluation of Technology Acceptance Model for Artificial Intelligence in E-
Commerce.” Heliyon 9(8):e18349. doi: 10.1016/J.HELIYON.2023.E18349.
24. Wang, Fan, Chunjiang Zhao, Hao Yang, Hongzhe Jiang, Long Li, and Guijun Yang. 2022. “Non-Destructive and in-Site Estimation of Apple
Quality and Maturity by Hyperspectral Imaging.” Computers and Electronics in Agriculture 195:106843. doi:
10.1016/J.COMPAG.2022.106843.
25. Wang, Yanpeng, Zhijun Zhang, Xingchen Wang, Xiao Yuan, Qian Wu, Shuaiyin Chen, Yangjun Zou, Fengwang Ma, and Chao Li. 2021.
“Exogenous Dopamine Improves Apple Fruit Quality via Increasing Flavonoids and Soluble Sugar Contents.” Scientia Horticulturae
280:109903. doi: 10.1016/J.SCIENTA.2021.109903.
26. Wójcik, Paweł, Jacek Filipczak, and Marzenia Wójcik. 2024. “Impact of Selenium Fertilisation of ‘Red Jonaprince’ Apple Trees on Selenium
Nutrition and Fruit Quality and Storability.” Scientia Horticulturae 327:112871. doi: 10.1016/J.SCIENTA.2024.112871.
27. Xia, Xue, Xiujuan Chai, Ning Zhang, and Tan Sun. 2021. “Visual Classification of Apple Bud-Types via Attention-Guided Data Enrichment
Network.” Computers and Electronics in Agriculture 191:106504. doi: 10.1016/J.COMPAG.2021.106504.
28. Zhang, Bo, Mengsheng Zhang, Maosheng Shen, Hao Li, Zhongxiong Zhang, Haihui Zhang, Zhaoyong Zhou, Xiaolin Ren, Yuduan Ding, Libo
Xing, and Juan Zhao. 2021. “Quality Monitoring Method for Apples of Different Maturity under Long-Term Cold Storage.” Infrared Physics
& Technology 112:103580. doi: 10.1016/J.INFRARED.2020.103580.
29. Zhao, Honglei, Ruotong Li, and Jiye Hu. 2023. “Frequently Used Pesticides and Their Metabolites Residues in Apple and Apple Juice from
Markets across China: Occurrence and Health Risk Assessment.” LWT 178:114610. doi: 10.1016/J.LWT.2023.114610.

View publication stats