Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 1 down vote favorite
1

Are there any multiple linear regression methods or packages that are resilient to occasional missing values? I have no prior view on imputing the missing values based on the nature of the data, and I would like to avoid discarding rows that have NAs.

Although I am not performing a panel regression, my data is arranged as panel data: Date, Identifier for individual in population, Characteristic 1, Characteristic 2 , ... , Objective function value.

share|improve this question
Define resilient. If the samples are iid and you missing only small portion of the data (and the missing values are distributed randomly), then linear regression can be thought as resilient to missing data. – mpiktas Mar 6 '12 at 4:06
If you don't want to discard the rows with NAs, and you don't want to impute missing values, what else is there to do? Or is that your question? – Peter Ellis Mar 6 '12 at 12:22
If the variables are categorical, then you can easily add "missing" as a category. But the right solution depends on the nature of the missingness: Are data missing completely at random (MCAR), missing at random (MAR) or nonignorable nonresponse? Also, what proportion of the data are missing? – Peter Flom Mar 6 '12 at 12:33
The data are missing at random. About 8% of the data are missing. The samples unfortunately are not i.i.d. – Quant Guy Mar 6 '12 at 14:24

Know someone who can answer? Share a link to this question via email, Google+, Twitter, or Facebook.

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.