A Two-Step Method of Statistical Difference Testing and K-Means Clustering for Identifying Quality Factors of Small Steel Bars

As profit of steel enterprises is getting smaller and comparison harder, in manufacturing industry quality inspection control have become main measures for over the difficulties. In sequential process, failure occurred preceding process make final worse. Detecting factor which affects to necessary stable efficient. Existing studies improve surface small bars assume that these been identified focus on diagnosis defects. However, there are no attempts validate based data. this paper, we attempt verify data using statistical data-mining techniques. To get practical problems from variation due operation date merge measurement quality, suggest method significance difference k-means clustering (k = 2). Method considers overall tendency about consider toward outlier. Quality factors bar can be detected by both methods serially. And apply real world case study.