Research On Classification Technology Of Breast Disease Data And Application Based On Attribute Reduction

With the development of computer technology, the speed of medical promotion of information technology is also speeding up. Promoting the using of information technology in medical reform is the direction of future development of the medical industry. At the same time, the most important task for medical data mining is using up the large amounts of medical data efficiently and play its potential role, and putting forward the corresponding scientific basis for decision making medical workers and researchers. The data attribute reduction is contracted to the particular application without losing data due value based on the original one. It should choose the smallest attribute subset can remove from the angle of the linear or nonlinear redundant and irrelevant attributes, and improve data accuracy and performance goals.Medical data is heterogeneous, massive, privacy, and data representation is not obvious.Therefore acquiring knowledge through appropriate data attributes contracted processing which is one of the important steps before using the method of data mining in medical data.Combination of data mining and data visualization can make users increase interaction before data mining in the process of data mining and after the data mining which is in order to improve researchers’ degree of comprehension.In this paper, starting from two directions which is the perspective of feature selection and feature extraction. Then applying interrelated algorithms for analyzing the two sets of medical data about breast disease. And using data visualization techniques to interpret and evaluate the mining objects, the mining tasks and the results efficiently. Then analyzing their interrelated performance indexes.Firstly, this paper briefly describes the background, significance and the current situation of domestic and international research. The classification and clustering of the data mining technology are also discussed. And it discusses the basic theories of the data mining technology and visualization. Then data mining result visualization, data mining process visualization and data visualization are expounded and illustrated. It also introduces two kinds of attribute reduction namely feature selection and feature extraction of data and their related algorithms.Secondly, this paper presents on a valid classification of a knowledge of breast cancer data mining by using the software of Weka. According to the data of breast cancer which is788 patients’ from a hospital in Lanzhou. Then choosing certain attributes from the breast cancer dataset using different types of feature extraction methods of data mining techniques.Then the classification algorithms are used to classify the data. The results show that, the accuracy using the algorithm of Bayesian network to classify is higher than other algorithms.The accuracy is improved obviously after using Logistic regression model algorithm for feature selection. The time taken to build model of classifying data subsets is obviously decreased, the time complexity has been reduced.Finally, the measurement value is greater than 0.9 according to testing the breast electrical impedance scanning detection data by KMO test. Therefore it is suitable for factor analysis. In this regard, it is proposed that classify the breast electrical impedance scanning detection data based on PCA algorithm effectively. Then using BP neural network to classify the data extracted from the source data to make diagnosis model by using of R language. The results show that the accuracy of classification based on PCA is not significantly lower, and the performance has obviously improved.