 after comprehensive discussion on store analysis, OLS analysis then we have already discussed about special analysis, special econometric, we have discussed about the partner data, we have discussed about the time-serve data. At this stage you have a clear idea that OLS is the best or SOAR is the best. when you have to use the OLS, when you have to use SOAR, when they are common, when they are uncommon, what are the characteristics of OLS, what are the characteristics of SOAR, what are the characteristics of your special analysis. when you do not correlate the error, there is no correlation, then the best approach is OLS. if you apply OLS on 5 models and correlate their error, then you have to move on SOAR analysis. so this is a very much simple approach. in the same pattern we have already discussed, if you have time series data and all the variables are integrated of order 1, the best approach is Johansson co-integration. if all the variables have mixed order of integration, then you can simply move on ARDL approach. and if all the variables are stationary at their level, they are stationary. and the best and efficient estimation technique is your OLS. in the same pattern, if your observation is not independent, they are geographically dependent, then you have to check the impact of geography and impact of neighbors. and to check the impact of a neighbor, you should scatter plot, you should do that graphical analysis, you should take the weight of the geography and see whether the coefficient of rho is equal to 0 or it is not equal to 0. in the same pattern, you have to check the correlation in your error term and if the error term are correlated, then SOAR is the best. and if they are uncorrelated, then OLS is the best estimator.