Research On Classification Algorithm Based On Fuzzy Rules

With the arrival of the information age,the data resources accumulated by all walks of life in daily management are increasing,and the emergence and growth of uncertain data and missing attribute value data bring great difficulties to data processing.So how to extract useful information from these data and classify it becomes critical.Fuzzy set theory is often used to deal with uncertainty problems,and data classification based on fuzzy knowledge is one of the important research contents of fuzzy set theory.The fuzzy classification rule is widely regarded as a good representation of classification knowledge,and it is readable and interpretable.Fuzzy classification is widely used in many fields such as image processing,character recognition,speech recognition,text classification,remote sensing,meteorological and industrial automation control.In recent years,the classification of uncertain data has attracted the attention of more and more researchers,and the traditional fuzzy rule classifier only considers single feature or considers multiple features.This paper presents a fuzzy rule classification algorithm based on association rule extraction algorithm,the basic idea of the classification algorithm based on fuzzy rules is to use a single feature first,and gradually increase the number of features for the samples divided into multiple classes and then reclassify it until there are no subdivision samples.In order to reduce the complexity of the classification algorithm,and improve the construction efficiency of classifier,we use a new calculation method of attribution degree: Calculate the ratio of the sum of the sample set membership grade of each rule to the number of rules,and take this ratio as the attribution degree of the class to verify that the process is approximately correct in the probability sense through the theoretical analysis.Finally,we test the performance of the classifier presented in the eleven public data sets,and compare with other six decision tree classifiers in accuracy.The test results show that the average classification accuracy in seven data sets of the classifier set up in this paper is higher than that of the other six classification methods.In order to deal with the classification problem of missing attribute values data,this paper presents a new data classification algorithm of missing attribute values based on the improved fuzzy feature extraction algorithm and the above fuzzy rule classification algorithm.The algorithm does not require interpolation operation of missing attribute values and it only uses the existing attribute values.The algorithm does not process missing items but directly makes the membership degree of the missing attribute value as 0,so that the effect of the uncertainty of interpolation on the classification is avoided.Four real data sets are selected from the UCI database,and this algorithm is compared with three existing algorithms for analysis.The test results show that the missing attribute value data classification algorithm based on fuzzy knowledge is improved in the missing data classification accuracy and generalization ability.