Design And Implement Of Business Management And Analysis System For Metrology Based On Data Mining

Traditional measuring business processing mass data in the face of many difficulties,unable to effectively extract the useful data and statistical data,so as to make the lack of data on the basis of business management and decision of measurement.Due to the customer,of measurement,the test items and the kinds of scale increasing and expanding,measuring business type and quantity of the data also will increase,for measuring the business data is particularly important for effective management.Data mining technology applied in the measurement business analysis and business management can improve the whole measurement inspection efficiency,favorable for the decision of metrology institute leader scientific support.First of all,according to the measurement supervision and inspection of tianjin academy of metrology business requirements and unit structure,analysis of system main functions,including measurement equipment,calibration,personnel,information,document information,testing data and other information management,data statistic analysis function provides leadership to check the measurement results with intuitive overall data support.The system management personnel can make convenient maintenance through system management for testing equipment,personnel and file information.According to the user department and the current user login time,automatically matching the user the instruments of point detection task,sending them to the user,information with active push model,by email or SMS inform inspector.Based on the above functional requirements,the system analysis,design,implementation,etc.Second,data warehouse construction.The system builds the data warehouse based on SQL SEVER to do the data mining.Ensure that the data is structured and provided to analysts in the most appropriate way,giving it a chance to fully understand the data.Then after you complete this model developed to standardize,using an iterative process development,adjust the network parameters and input level to achieve an acceptable range,or until the error model shows that there is no room for further improve.Finally,the prediction model based on k-means algorithm and collaborative filtering algorithm was completed,and the parallelization of k-means was implemented based on the hadoop platform.The improved collaborative filtering algorithm is based on the comparison between enterprise users and the recommended results based on the test project.The optimal scheme is recommended based on the detection project.Conduct classification analysis of test items and corporate customers.Detection project classification analysis instruments makes the basic classification,first of all instruments can be divided into basic K class,as clustering number of Kmeans clustering algorithm,and then initialize clustering center K,corporate clients after inspection instruments of testing data and the K class,of which the most close to the equipment assigned to the corresponding class.Based on the Kmeans algorithm,the customer can complete the classification analysis based on the area of the inspection enterprise,the type of inspection,the time of delivery and so on.The improved collaborative filtering algorithm is introduced.Based on the type of appliance that customer A similar to customer A,estimate the type of appliance that client A may be sent to check.Metrology institute can provide better personalized service,access to send the goods with the customer types and the changing rule of the factors such as region,found the regularity of measuring business behind testing project data changes and potential information.The senior leadership can view the classification analysis of the inspection users and the characteristic analysis of the product,so as to make a reasonable plan for the future work.