In statistics, linear regression refers to an approach for modeling the relationship between a scalar response variable and one or more explanatory variables.
a) How would you construct a small data set (less than 50 data points) with 2 explanatory variables, such that both explanatory variables and the response variable are Gaussian, but the residual of the linear regression is not Gaussian?
b) How would you construct a small data set (less than 50 data points) with 2 explanatory variables, such that neither the explanatory variables nor the response variable is Gaussian, but the residual of the linear regression is Gaussian?
programming-solved!
This is a good question. Check out this post, it should help explain why this is case. I can't really explain this concept much better than the people here did.
http://stats.stackexchange.com/questions/12262/what-if-residuals-are-no…
Thank you for the useful information!
I think (a) is impossible (since the sum of gaussians is gaussian) and (b) is trivial - make X and Y be your two explanatory variables with U(-1,1) distributions and make Z (response variable) = X + Y + G (where G is a gaussian deviate with 0 mean and 1 std dev)
Then again it's been a while since I did real quant stuff so I may be missing something obvious here.
Assumenda quis qui quis quia. Adipisci veritatis nulla aliquid id a voluptatem. Assumenda voluptatum aut eum minus ut ducimus fugiat. Labore nobis reprehenderit sunt harum incidunt nesciunt dolores. Natus qui eos quidem in. Quaerat explicabo voluptatem accusamus aspernatur ut officia labore.
Voluptatem maxime qui id quia voluptatem. Assumenda placeat corporis deserunt suscipit consequuntur minima et.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...