R-squared is used for linear regression analysis by transforming the dependent variable in a regression model that is fitted to a particular set of data. In order to use R-squared on a specific regression model, a statistician may have to start with other transformations such as logging and deflating so as to eliminate unnecessary variability caused by factors such as inflation. These transformations explain a significant amount of variance.
The figures generated by the deflation transformation are explained through the use of other models because they may differ greatly with the actual variance. The random walk and the random trend models are typically fitted to the difference in the data to reduce the first and seasonal differences in the variance. Applying these models, their mean squared error can be obtained as it is the same as the variance of the differenced series of data. However, it is advisable to calculate an "effective" R-squared value using the deflated series above. The R-squared value is then calculated by taking the mean squared error, dividing it by the variance, and finding the percentage of the result subtracted from 1. There is no definite figure that should be used for R-squared in regression analysis, but the above steps are crucial for most regression models.