同欣電子是一間專門製造印刷電路板的台灣公司。在他們的生產線上，會產生很多的瑕疵品。其中一種瑕疵品叫導體滲鍍。到目前為止，該公司並不了解造成該瑕疵品的可能因素。透過分析生產資料，我們希望能獲得最有可能造成該瑕疵品的站別和生產參數。然而，紀錄這種瑕疵品的生產資料量太小，導致任何的機器學習模型都無法正確地解決問題。儘管如此，我們仍然可以使用統計方法分析生產資料來獲得結果。基於分析結果，公司可以更謹慎地調整他們的機器參數，進而降低造成該瑕疵品的機率，在未來同時也可以提高生產良率。;Tong Hsing Electronics Industry is a Taiwan-based company that specialized in PCB manufacturing. Currently, they still found many bad products from their production line. One type of defect that they encounter is “Bridging Conductor”. Until now the company doesn’t have any strong knowledge about possible factors causing that defect to their products. By analyzing the production line log data hopefully, we can obtain station and production parameters that most likely contributing to the defect. However, the production line log data for this type of defect are too small to be analyzed, so that any machine learning models can’t be used to solve the problem correctly. Nevertheless, we still can analyze the data by using any statistical approach to obtain the result. Based on that, the company can adjust its machine parameters more carefully, thus decreasing the probability of finding these defects while also increasing its production yield rate in the future.
