search

mkao006 / sws_imputation

Repository for implementing imputation for the Statistical Working System
0 stars 0 forks source link

Considering log the variables #51

Closed mkao006 closed 10 years ago

mkao006 commented 10 years ago

Logging the data allows the problem to be solved in an additive way and also prevent negative imputation of values.

mkao006 commented 10 years ago

Add in an option to log the yield in meanlme4.R and swsImputation.R

mkao006 commented 10 years ago

Logging the variable seems to amplify the variance of the coefficient estimated for the regional mean. Some imputed value becomes extremely large after the exponential transformation, if we remove the regional mean variable then the fit is no better than the un-logged model.