Data Mining Technique in Wafer Process Alarm Data Analysis

• Main Authors: Yu-Chun Huang, 黃玉君
• Other Authors: Ben-Chang Shia
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/15411001646415671238

碩士 === 輔仁大學 === 應用統計學研究所 === 97 === In the semiconductor manufacturing process, because of the cumbersome process and a large amount of data, and the manufacturing process with the development of new products with each passing day, leading to not be able to do data analysis, the majority of parameters are set based on engineer’s know-how and experience ,not there is a real reference, however, when personnel changes, the experience will be with the personnel changes and the passage of time. Therefore, the main propose of this study was to apply to the semiconductor wafer THP machine based on the value of the parameters set by the descriptive statistics, dissemination of maps and data mining of Logistic regression for parameter setting model of THP. The use of data mining and statistical analysis, the regression equation to identify the THP can be found on the law of value must exist, will be subject to process, brands, products and other related impact analysis of the results of this study was to establish the model, will allow early warning determine the hit rate was 69.5%, sensitivity of up to 66.67 percent, after discussions with the engineers found that with the current situation and to be aware of the usefulness of this model, engineers can give online as a future basis for setting parameters.