The preceding statements initially define LogY to be a vector of missing values. Idx = loc(Y > 0) /* find indices where Y > 0 */ LogY = j(nrow(Y),1.) /* allocate missing */ The following example uses b=1 and calls the LOG10 function, but you can call LOG, the natural logarithm function, if you prefer. In the SAS/IML language, this transformation is easily programmed in a single statement. For the latter choice, you can show that a = b – min( Y), where b is either a small number or is 1.
Some people like to choose a so that min( Y+a) is a very small positive number (like 0.001). The transformation is therefore log( Y+a) where a is the constant. How do you handle negative values if you want to log-transform the data?Ī common technique for handling negative values is to add a constant value to the data prior to applying the log transform. However, some quantities (for example, profit) might contain a few negative values. In many cases, the variable of interest is positive and the log transformation is immediately applicable. (Remember, however, that you do not have to transform variables! Some people mistakenly believe that linear regression requires normally distributed variables. Common examples include data on income, revenue, populations of cities, sizes of things, weights of things, and so forth. A log transformation is often used as part of exploratory data analysis in order to visualize (and later model) data that ranges over several orders of magnitude. It is used as a transformation to normality and as a variance stabilizing transformation. Thus they are no longer comparable with the GLS estimated results.The log transformation is one of the most useful transformations in data analysis. The presence of SIGMASQ changes the coefficient of my independent variables and the DW value. However I would like to get rid of SIGMASQ if it is possible. In the new versions of EVIEWS 9 and 10, when I add an AR(1) to my independent variables, I get an extra independent variable in the regression printout named SIGMASQ. The same value for the corrected coefficient and the improved DW, I was getting it by simply estimating the GLS by transforming the variables and using the estimate of Rho from the DW value from the original regression which had presence of serial correlation. This was correcting the DW to be closer to 2 and a more correct slope coefficient. I did this by adding an AR(1) term as an additional "independent" variable when I ran the regression. With EVIEWS 8 and below I used to estimate an equations in which I tried to correct for the presence of serial correlation.
0 kommentar(er)
