#1
Hello there,

I have data set of 5846 observations out of which 15 observations are outliers. I need to perform multiple linear regression which as I know is highly sensitive to outliers.

Do I have to filter those 15 outliers out or they will not mess up my analysis, since the number of outliers is very small in comparison to other observation??

Please Help!! :wave: :wave: :wave:
 

Miner

TS Contributor
#2
Be very careful in dealing with outliers. If they are obvious data entry errors, it is okay to remove them, but often they may be indicative of a lurking variable that you did not consider in the data collection. You can use robust methods or try your regression with and without the outliers to see whether they impact your results/conclusions.