Hellow! can you please send me the value of F (SE)?, because I think if you want your statistical analysis significant ... you have to obtain F (SE) under the value 0.01(P).
You need to describe what you have much more clearly than you have done. I believe that you are saying that you ran a regression with R^2 = 0.9 and found a pattern in the residual plot. If that is the case, you have more work to do to fix the problem in the residual plot. See Kutner, et al, Applied Linear Statistical Models, 5th ed, McGraw Hill/Irwin or an equivalent reference. Best wishes.
I assume the same thing as David ("You ran a regresson with R^2 = 0,9 and found a pattern in the residual plot").
It happens, when you fit a regression model with very high R-squared but still you have found some pattern in the residual plots.
High R-squared means that your fitted values are very close to observed values, i.e, the regression line explains quite much variability in your response data. But one of the limitation of R-squared is that: "R-squared cannot determine whether the coefficient estimates and predictions are biased, which is why you must assess the residaul plots ". Beside the reference refered by David, here is a link to give you a quick discussion:
http://blog.minitab.com/blog/adventures-in-statistics/regression-analysis-how-do-i-interpret-r-squared-and-assess-the-goodness-of-fit
It means that, before using R-squared to say something about your model, first you need to check if your model assumptions are satisified. By looking at the residual plots, and use some other tests, for the assumption of independence, constant varince, randomness...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics