Comparative Evaluation of Machine Learning Models for Predicting Air Permeability of Knitted Fabrics

Öz

Air permeability is a key property of knitted fabrics that directly influences thermal and moisture comfort, particularly in sportswear, innerwear, and outerwear applications. In this study, air permeability values of 21 knitted fabric samples—differing in yarn count, twist level, knit structure, and fabric density—were used to develop predictive models using various machine learning algorithms. The models tested include linear regression (LR), artificial neural networks based on the Multilayer Perceptron (MLP) architecture (MLP), support vector machines regression (SMOreg), random forests (RF), and multivariate adaptive regression splines (MARS). All models were evaluated using 10-fold cross-validation, and their generalization ability was assessed on independent validation data. Among these, the support vector machine regression (SMOreg) model achieved the highest predictive accuracy, with low error rates and strong correlation coefficients. In contrast, the MARS model performed poorly due to its limited capacity to capture structural variability in small datasets. These findings highlight the effectiveness of AI-based modeling in predicting fabric performance and contribute to the advancement of data-driven practices in textile engineering.

Anahtar Kelimeler

• Knitted fabric
• Air permeability
• Machine learning
• Support vector machines
• Artificial neural network

Örme Kumaşların Hava Geçirgenliğinin Tahmininde Makine Öğrenmesi Modellerinin Karşılaştırmalı Değerlendirilmesi

Öz

Hava geçirgenliği, özellikle spor giyim, iç giyim ve dış giyim uygulamalarında termal ve nem konforunu doğrudan etkileyen örme kumaşların temel bir özelliğidir. Bu çalışmada, iplik numarası, büküm düzeyi, örme yapısı ve kumaş yoğunluğu gibi farklılıklar gösteren 21 örme kumaş numunesinin hava geçirgenliği değerleri kullanılarak farklı makine öğrenmesi algoritmaları ile tahmin modelleri geliştirilmiştir. İncelenen modeller arasında doğrusal regresyon (LR), yapay sinir ağlarının bir alt modeli olan çok katmanlı algılayıcı (MLP), destek vektör makineleri regresyonu (SMOreg), rastgele ormanlar (RF) ve çok değişkenli uyarlanabilir regresyon eğrileri (MARS) yer almaktadır. Tüm modeller 10 katlı çapraz doğrulama ile değerlendirilmiş ve genel bağlantı yetenekleri bağımsız doğrulama verileri üzerinde test edilmiştir. Bu modeller arasında, destek vektör makineleri regresyonu (SMOreg) modeli en yüksek tahmin doğruluğunu elde etmiş, düşük hata oranları ve yüksek korelasyon katsayıları göstermiştir. Buna karşılık, MARS modeli küçük veri kümelerindeki yapısal değişkenliği yeterince yakalayamaması nedeniyle düşük doğrulukta tahminler üretmiş ve yüksek hata oranlarıyla en başarısız model olarak değerlendirilmiştir.

Anahtar Kelimeler

• Örme kumaş
• hava geçirgenliği
• makine öğrenmesi
• destek vektör makineleri
• yapay sinir ağları

Kaynakça

1. [1] Li, Y., ‘‘The science of clothing comfort,’’Textile Progress, 31 (1-2): 1-135, (2001).
2. [2] Mavruz, S., & Oğulata, R. T.,‘‘Investigation and statistical estimation of air permeability in cotton knitted fabrics.’’Textile and Apparel, 19(1): 29-38, (2009).
3. [3] Luo, X., Hou, W., Li, Y., & Wang, Z.,‘‘A fuzzy neural network model for predicting clothing thermal comfort,’’. Computers & Mathematics with Applications, 53 (12): 1840-1846, (2007).
4. [4] Fayala, F., Alibi, H., Benltoufa, S., & Jemni, A., ‘‘Neural network for predicting thermal conductivity of knit materials,’’Journal of Engineered Fibers and Fabrics, 3(4):53-60, (2008).
5. [5] Siddiqui, M. O. R., Ali, M., Zubair, M., & Sun, D., ‘‘Prediction of air permeability of knitted fabric by using computational method,’’Textile and Apparel, 28(4):273-279, (2018).
6. [6] Daukantiene, V., & Vadeike, G., ‘‘Evaluation of the air permeability of elastic knitted fabrics and their assemblies,’’International Journal of Clothing Science and Technology, 30(6):839-853, (2018).
7. [7] Demiroz Gun, A., Unal, C., & Unal, B. T.,‘‘Dimensional and physical properties of plain knitted fabrics made from 50/50 bamboo/cotton blended yarns,’’Fibers and Polymers, 9:588-592, (2008).
8. [8] Unal, P. G., Üreyen, M. E., & Mecit, D., ‘‘Predicting properties of single jersey fabrics using regression and artificial neural network models,’’Fibers and Polymers, 13: 87-95, (2012).

9. [9] Nabiyev, V.,‘‘Artificial intelligence,’’Seckin Publishing, Ankara, (2016).
10. [10] Kaluza, B., ‘‘Machine learning in Java,’’Packt Publishing, UK, (2016).
11. [11] Torkul, O., Gülseçen, S., Çağıl, G., & Uçar, M.,‘‘Artificial intelligence and applications in engineering.’’Sakarya University Publications, Sakarya, (2017).
12. [12] Geze, R. A., & Akbaş, A., ‘‘Detection and classification of fabric defects using deep learning algorithms,’’Politeknik Dergisi, 27(1): 371-378, (2024).
13. [13] Bayram, B. S., Yıldız, O., & Korkut, İ., ‘‘Predicting surface roughness with machine learning’’.Politeknik Dergisi, 28(1):187-195, (2024).
14. [14] Aydın, N., Erdem, O. A., & Tekerek, A., ‘‘Comparative analysis of traditional machine learning and transformer-based deep learning models for text classification.’’Politeknik Dergisi, 28(2), 445–452, (2025).
15. [15] Akkis, B., ‘‘The effect of different weave types knitted from different yarn counts on the physical properties of fabric,’’Master's thesis. Çukurova University, Institute of Science and Technology, Adana, (2009).
16. [16] Aydın, S., & Özkul, E., ‘‘Data mining and an application in Anadolu University open education system,’’Journal of Education and Training Research, 4(3), (2015).
17. [17] Kızılkaya, M. Y., & Oğuzlar, A.,‘‘ With R Programming, Comparison of Performance of Different Machine Learning Algorithms,’’Karadeniz International Scientific Journal, 3(2): pp. 90-98, (2018).
18. [18] Aydin, D. (2014). ‘‘Applied regression analysis,’’ Nobel Publishing, Ankara, (2014).
19. [19] Kargı, V., ‘‘Artificial neural network models and application in a textile company,’’Ekin Publishing, Bursa, (2015).
20. [20] Shamey, R., & Hussain, T, ‘‘Artificial intelligence in the color and textile industry,’’Journal of Color, vol. 33(1): 33-45, (2003).
21. [21] Sevimli, Y., ‘‘Application of multivariate applicable regression extensions in a split-mouth study,’’Master's thesis, Marmara University, Department of Biostatistics, Istanbul, (2009).
22. [22] Ceyhun, G., ‘‘Comparison of the performances of data mining methods used in classification according to data set characteristics,’’Doctoral dissertation, Gazi University, Institute of Educational Sciences, Ankara, (2020).
23. [23] Li, C., & Jiang, L., ‘‘Using locally weighted learning to improve SMOreg for regression,’’. In 9th Pacific Rim International Conference on Artificial Intelligence Proceedings, China, (2006).
24. [24] Aydemir, E., ‘‘Artificial intelligence with Weka,’’Seçkin Publishing, Ankara, (2019).