Research And Application Of Sensitivity Analysis Methods Based On Quantitative Association Rules

With the development and popularity of Internet,the rapid increase of data make the cost of getting information less and less.However,it is more and more difficult to dig out the relevant information in the big data.The data sensitivity analysis provides a data analysis method in the discovery of valid information,which helps people find effective information.And it has been applied in various fields and has become a hot spot of current research.The most critical technology of sensitivity analysis is to describe the degree of importance of the input variables to the output variables.However it is inefficient and it is difficult to prove the reasonableness.Data mining technology is to find the relationship between the attributes of data set in a large amount of data.And the association rules mining technology is one of the main analysis methods.Association rule mining has been widely used,but it generates a lot of uninteresting rules and cannot dig out the sensitivity between the elements.Therefore,with the numerical attribute value replaced by its change proportion,the association rule can reflect the correlation between the degree of change of attribute value to analyze the degree of importance of the mutual influence and the correlation between attributes,and to realize the sensitivity analysis,which is a practical value of the new method of data analysis.In this paper,a method of the sensitivity analysis based on quantitative association rules is defined to find out association rules of the high degree of mutual impact by SA-Apriori algorithm.First,it uses the relative value to describe the degree of numerical attributes change and their impact on the target variable change.Second,it completes discrete of relative value based on the same width partition.Based on this method,the relative score concept is proposed and two different sensitivity interest indexes are established.The smaller the range of change for one or more attributes,but the greater the range of influence on another or several target attributes,the more sensitive it is.Second,according to the method of analysis of sensitivity based on quantitative association rules,the sensitivity analysis software was programmed in QT platform,and the design and implementation of core module are introduced in detail.Third,with student test results of data structure of a university recounted in accordance with the autonomy of the knowledge points,the data structure of each student in a course of knowledge points can be analyzed based on the different data size,different relative score concept and the two different sensitivity interest indexes.Experiments show that the algorithm can achieve the sensitivity analysis.It is not only found that the sensitivity rules are used in different granularity and the granularity of the data with the knowledge block is the best.Choosing the concept of different relative scores is based on the.And selecting the different sensitivity of interest indicators is related to the mining rules.