I find Durbin-Watson value 0.35 with high multicorrelation coefficient equal to 75%. However, when I transform the variables with FD transform method the Durbin-Watson values rises to 1.7, whereas multicorrelation coefficient values comes down to 37%. although the CC is significant but the proportion explained by the independent variable becomes very low. When I use CO Transform method I do not find improvement in the Durbin-Watson value, but the CC remains around 73%. I do not understand what to do now. Please suggest. My data is time series data on visibility, RH and some chemical constituents in the atmospheric aerosol. I wish to develop a regression equation to see the impact of meteorological and chemical parameters on the visibility.
