Garment Time Prediction of Denim Products using Artificial Neural Networks and Ensemble Methods

Authors : Yusuf Kuvvetli, Ebru Çalışkan

Pages : 220-247

Doi:10.51541/nicel.1802023

View : 87 | Download : 118

Publication Date : 2025-12-31

Article Type : Research Paper

Abstract :Textile products are subject to frequent changes due to fashion trends, yet maintaining low production costs remains essential. Production time affects these costs significantly and numerous factors can affect production time, including the complexity of the design, the type of fabric, specific parts, and operational characteristics. This study aims to predict the garment time of denim products using information from sample items, rather than relying on traditional time measurement methods. To achieve this, a dataset of 79 attributes was created and later refined to 53 features through preprocessing. These selected features were then used to predict the standard garment time using artificial neural networks, extreme gradient boosting, and random forest algorithms. Based on these results, the study proposes an alternative approach for managers to predict production time using sample-based parameters, eliminating the need for time-consuming observational studies. The findings indicate that ensemble algorithms such as extreme gradient boosting, and random forest demonstrate a high degree of efficacy, exhibiting an absolute average error of approximately 10% in the test data.
Keywords : Zaman etüdü, Konfeksiyon, Standart süre, Denim ürünleri, Makine öğrenmesi

ORIGINAL ARTICLE URL