标题: 用SAS将非正态分布数据转换为正态分布数据，怎样实现？谢谢 [打印本页] 作者: shiyiming 时间: 2006-4-19 15:37 标题: 用SAS将非正态分布数据转换为正态分布数据，怎样实现？谢谢 用normal检验了我的数据是非正态分布的，请问如何才能将他们转换为正态分布？谢谢作者: shiyiming 时间: 2006-4-20 09:33 标题: 想怎么弄就怎么弄. 要转换为正态分布还不容易吗?把样本曲线与正态分布曲线比较,多退少补,不就满足你的要求了吗?作者: shiyiming 时间: 2006-5-10 03:08 标题: Use log transformation A lot of people use log transformation, it works for most kinds of data. just create a new variable in your data step,
NewY=log(y);
and do Proc Univariate to test your new variable called 'NewY', if it still not normal, try other mathmethical functions like 1/y, sqrt(y), exp(y),...., or convert your data into categorical data like 'very low, low, medium, high, very high' and use categorical data analysis like Proc freq, proc logistic, proc genmod.
Or use your original data, and choose more robust procedures to do your statistical tests, like PROC NPAR1WAY, PROC NLMIXED, PROC GLIMMIX.