相关系数的不确定度分析及其应用

目的:探讨表示因子间相关性的相关系数的可靠性。方法:应用集对分析理论,将应变量X与自变量Y进行同异反分析,建立X、Y的联系度μ,从而将r分解为ra(或rc)部分的确定相关和rb部分的不确定相关。以相关系数的不确定度f s=rb/r=b表示相关系数的可靠性。结果:新昌县脑卒中发生人数与气压、气温的相关分析表明,脑卒中发生人数与日平均气压、最高气压、日平均气温、最低气温、气温日较差、前后两天气温差绝对值、前后两天气压差绝对值、气压日较差的相关系数均通过0.05的显著性检验水平,但只有后三者相关系数的不确定度较低,有较高的可靠性,和相关研究结果一致。结论:相关分析时,只有当相关系数既达到显著水平,不确定度又较低,相关关系才真实可信。

Analysis of Uncertainty in Correlation Coefficient with Set Pair Analysis

Objective: Discussing reliability of correlation coefficient.Methods: The dependent variable X and independent variable Y were analysised with method identical-discrepant-contrary,and r was decomposed ra(or rc)and rb.Uncertainty in Correlation coefficient(fs=rb/r=b) figured the reliability of correlation coefficient.Results: The result of correlation analysis for cerebral apoplexy and air pressure,air temperature showed : the correlation coefficient of cerebral apoplexy and difference in temperature,difference in pressure,daily range of pressure had high reliability.Conclusion: Only correlation coefficient achieved significance level and had low degree of uncertainty,the correlativity was reliability.