Post Your Answer

Yes you heard it right. There is an unusual observation in regression diagnostics. The way through which it can be unusual is outliers. An observation with large residuals are known as outliers. You can also consider it as an observation in which the dependent-variable is unusual. An outlier usually indicates a data entry error.

Ramesh, not just outliers, there is another way that an observation can be unusual. This is called influence. An observation is influential if by removing the observation, there are changes in the estimate of coefficients.

Is there any other way that an observation can be unusual?

Yes there is. https://stats.idre.ucla.edu/sas/webbooks/reg/chapter2/regressionwith-saschapter-2-regression-diagnostics/ This site has all necessary information