Association Rules Mining Based On Correlation Analysis And Its Application

Association rule mining is an algorithm for mining and extracting the existing relationship in a transaction project from a database with massive data.With the emergence of a series of problems such as fraud identification,business decisionmaking,fault analysis and disease detection,the rapid development of association rule mining algorithms has been promoted.The electronic communications industry and banks have an urgent need to identify fraud;In business,when making business decisions,it is important to use algorithms to select a large number of uncertain factors to help users make quick decisions;In the process of traditional industrial failure analysis,the staff can only perform fault detection based on intuition and own work experience.Faced with numerous production features,it is easy to miss detection,which greatly reduces the efficiency and speed of detection;for medical staff,the early symptoms of major diseases are often mild,and it is easy to miss the diagnosis and misdiagnosis.When patients have obvious symptoms,most of them are more difficult to cure.In fraud identification and business decision-making,there are already mature methods to solve the above problems.This paper studies and analyzes fault analysis and disease detection based on association rules mining.Firstly,this paper studies the fault analysis based on association rules mining.Aiming at the problems in industrial production,a correlation analysis model was constructed.The model used Lasso regression algorithm and six statistical methods(F test,chi-square test,decision coefficient,correlation coefficient,error variance,and wilks lambda value).The model extracts stamping bed data and quality data from an automotive plant for analysis.First,the regression algorithm is used to sparsely process industrial data.Second,six statistical methods are used to calculate the correlation for the sparse data.Third,the results are integrated and analyzed to find out the factors that have a greater impact on the final pass rate in the factory.Finally,we apply the correlation analysis model to the actual generation and verify it by professionals in the factory: In the event of a fault,the model can quickly diagnose the characteristics of the correlation with the faulty station,thus helping the professional to quickly adjust,greatly saving the time cost of fault diagnosis.Secondly,this paper studies the disease detection problem based on association rules mining.At present,the early diagnosis and treatment of major diseases are diagnosed only by the experience of doctors,and there is often a series of uncertainties.Therefore,an association rule mining model is constructed,which uses MDLP(entropy-based minimum description length criterion)discretization algorithm and Apriori algorithm.The model selected five physiological signals(heart rate,pulse,blood pressure,respiratory frequency,blood oxygen concentration)of 17 patients with brain damage,heart failure and sepsis.First,using the MDLP(entropy-based minimum description length criterion)discretization method with better discretization results,and the more common equal-width discretization algorithm for comparative analysis.Second,the discretized results are mined using the Apriori algorithm,third,the correlation between the range of physiological signal eigenvalues and some major disease quality tests is analyzed.At last,the association rule mining model is used to help medical staff determine the disease according to the change of physiological signal characteristic value in the early stage of the disease,so as to carry out key diagnosis and treatment,which reduces the possibility of missed diagnosis to some extent.In summary,I use the association rule mining algorithm for fault analysis and disease detection,construct a correlation analysis model and association rule mining model,and verify the validity of the model through real experimental data,which has deep practical application value.