Automatic prediction and cross validation of textile fabrics data using machine learning methods

The principles of machine learning methods have been used for classification and prediction of 105 sample fabrics. The secondary database was collected from Textile Department, Chennai. Physical properties of fabrics were measured using the well known Kawabata Instrument. Fabric properties such as Tensile Energy, Tensile Resilience, Bending Properties, Shear properties, Compression Properties and Surface Properties have been measured on fabric samples and used in the present study. Initially, different methods of classification have been constructed for three known categories of fabrics made up of Polyester, Lyocell/Viscose and TreatedPolyester. The classification yielded cent percent prediction and accuracy. The application of SVM, RF, kNN and Logistic Regression test score, prediction and confusion matrix validated the grouping in case of individual category of fabrics as well as combined group. The categorization of fabrics is very essential for textile industry and other fields also.