Data Analysis Study For Blending Products

In recent years, data analysis has got extensive and satisfactory applications in many fields, such as finance, retail, telecommunication, biology, network intrusion-detection systems and so on. Accordingly, the data analysis tools are becoming mature, for example, the SPSS in statistical analysis and the WEKA in machine learning are more friendly, and Microsoft has embedded module for data analysis into its SQL 2005 recently. Of course, there are many other data analysis tools used in special fields. Under this trend, many corporations begin to make use of their data which has been accumulated many years for data analysis relative to their products and the production.This paper is based on the data analysis for blending product design, craftwork parameters and quality control in blending vocations for many years, and has summarized a suitable processing that applies to data analysis for blending products, and has also introduced many application points and measures of data analysis for it. Recently, the requisite points of data analysis for blending products(including drinking, delicatessen, tobacco, liquor, Chinese traditional medicine and so on) concentrate in difference analysis , relation analysis ,and forecast analysis. The difference analysis deals with area, year, part of raw materials. The relation analysis deals with the relationship between inherent physics and chemistry elements of raw materials, the relationship between raw materials and auxiliary materials, the relationship between craftwork parameters and product quality. The forecast analysis deals with rang estimation and the relationship between product quality property and product control parameter. Thus, there are many methods involving in this kind of data analysis, including variable analysis, square difference, stepwise linear regression, principal components analysis, path analysis,scatter plot,box plot in statistics, and M5P decision tree, ERP forecast tree in machine learning field, and RBF network, BP network in computational intelligence field. This paper elaborates some major methods emphasizing particularly on the theory and application, because different data analysis methods apply to different objects, and some data relates to the corporation secrets, thus, here shows partly application results of some major methods. This paper is a summarization of the corporation application item and long-term practice, and is worthy for the research and practice of data analysis for the vocation involving blending product.