Design And Implementation Of Analytical Algorithms In The Field Of Computing Engine Testing

With the rapid development of the field of big data in my country,the calculation engine tools generated for big data calculations are overwhelming,but there are almost no tools to test the calculation engine.Therefore,this paper selects the current mainstream computing engine for testing in both offline and real-time parts.In addition,this article applies machine learning algorithms to the testing field for the first time.After obtaining the corresponding test items,this article firstly uses SVM,KNN,and improved KNN to analyze the test items under offline data;then,in offline and real-time,it uses SVM,decision tree,and Relief F?decision Tree algorithm to analyze the test items.Finally,the optimal model was selected.This article first elaborates the research background and relevant significance of computing engine and testing work,and introduces the relevant research status.Secondly,it summarizes the related theories of the current popular computing engines,the technical overview of the test data sets involved in the test computing engines,the steps of the classification algorithm,and the evaluation indicators of the classification algorithm.This article introduces the data set from three aspects: data source,data feature selection.First,it studies the basic theories and advantages and disadvantages of SVM,KNN,and decision trees.The KNN algorithm is improved by the weighted value,and the weighted-KNN algorithm is introduced,and a comparative experiment is carried out on the offline engine test item data set.Secondly,using the Relief F algorithm to improve the decision Tree,I propose the Relief F?decision Tree algorithm,and conduct comparative experiments on the offline and real-time engine test item data sets.Finally,the best model was selected.Finally,based on the study of the previous chapters,this article implements two test tools,namely offline computing engine testing and real-time computing engine.After using automated testing technology to execute the designed test cases for many times,and I manually sort out test sets for test items.Then,it selects the appropriate model for analysis.Finally,it displays the system interface.Experiments have proved that the theoretical method proposed in this paper is effective and feasible.It has designed two major testing tools.Although the selected computing engine is Spark,it can also test other computing engines.Secondly,the test item analysis algorithm it has made for the field of computing engine is a great reference for the test work and even the entire test field.