Abstract:

Purpose- In this study, textile goods, of manufactured by a textile company, whether being returned because of defects or not has been investigated by data mining and machine learning techniques. Main purpose of the study is to determine which of the products passing through 15 different production lines during the manufacturing process being defective and faulty at the last stage. Methodology- In this study, there are 250 different variables and 72959 lines of data on the production line. In order to perform a data mining process, it is firstly necessary to understand the data and determine the process. For this, CRISP-DM algorithm has been used. Modelling and classification algorithms are applied to estimate the production of faulty goods. In the model, a supervised learning model based machine learning methods have been used. The dimensions, loops and some statistical features of the data have been examined, and then it has been studied in the Python programming language. The feasibility of model and success rates have been evaluated with findings. Findings- The results of the model show that logistic regression and k-nearest neighbour algorithms give above %90 percentage confusion rate. It has been said that with this model is succesful for predicting defective and faulty product in manufacturing line. Conclusion- It has been tried to predict whether there will be faulty products that reduce quality. With this study, it has been aimed to give a signal to the production line in advance.

Keywords: