https://communities.sas.com/t5/Statistical-Procedures/log-transformation/m-p/710554#M34386 <P>Hi.</P><P>&nbsp;</P><P>How can do a log transformation of a variable if the distribution is not symmetric. Is there a proc for that?</P><P>&nbsp;</P><P>Thanks, Asif&nbsp;&nbsp;</P> Mon, 11 Jan 2021 14:36:41 GMT asifgeneral 2021-01-11T14:36:41Z https://communities.sas.com/t5/Statistical-Procedures/log-transformation/m-p/710554#M34386 <P>Hi.</P><P>&nbsp;</P><P>How can do a log transformation of a variable if the distribution is not symmetric. Is there a proc for that?</P><P>&nbsp;</P><P>Thanks, Asif&nbsp;&nbsp;</P> Mon, 11 Jan 2021 14:36:41 GMT https://communities.sas.com/t5/Statistical-Procedures/log-transformation/m-p/710554#M34386 asifgeneral 2021-01-11T14:36:41Z https://communities.sas.com/t5/Statistical-Procedures/log-transformation/m-p/710564#M34387 <P>You may want to describe more of what you are doing and where you need a transform, especially since you question references symmetry of distribution.</P> <P>&nbsp;</P> <P>The basic data step function Log will transform values in the appropriate domain, best creating a new variable.</P> <P>&nbsp;</P> <PRE>data want; set have; logvariable = log(variable); run;</PRE> <P>&nbsp;</P> Mon, 11 Jan 2021 15:02:26 GMT https://communities.sas.com/t5/Statistical-Procedures/log-transformation/m-p/710564#M34387 ballardw 2021-01-11T15:02:26Z