Research On Correlation Analysis Of Data About Drug Abuse And The Related Epidemic In Yunnan Rovince

In recent years, drug abuse situation becomes more and more serious. By the end of2014, there were186000drug users in Yunnan province. According to the ministry of public security, a survey of Yunnan drug users explicit and implicit than at gezer. Yunnan province is facing a serious crisis of drug abuse. Drug abuse has triggered the series of epidemic hazards in our daily life.In this paper, used the traditional method of mathematical statistics combined with modern data processing technology, applied to the data of drug abused and related epidemic of Yunnan province, the data correlation analysis and method of research.First, this thesis investigated from2004to2013, voluntarily to a center for disease control and prevention in Yunnan province for2002samples data of methadone drug treatment. Statistical analysis of drug users in Yunnan region alone infection and cross infection of HIV, HCV, TB epidemic situation.Then, in this paper, the data of the univariate analysis, using chi-square test to identify individual factors associated with epidemic, confounding bias method to find related theoretical results and the actual circumstance of deviation between variables. Univariate analysis results, the use of binary logistic regression analysis to correlation analysis of data, find out the best description of epidemic situation associated variables. We found that HCV test consequences have high correlations between given drugs, HIV test results, with no treatments of illicit drugs, HIV test results and HCV test results, TB test results, half illicit drugs.Finally, in this paper, correlation analysis of drug abusing and epidemic promotion methods. Due to logistic regression analysis will all data as the training data, to test, the conclusion of this article constructs the correlation between variables of BP neural network model for testing; For multiple possible redundancy associated variables, this article introduce the KPCA to improve logistic regression analysis.