calculate the mean with outliers

#1
Hello, I'm trying to calculate the mean of a list of about 1000 values. These values are times of CPU usage, the problem is that because of errors in measurement I've some outliers. If I calculate the mean of all these values the outliers mask the good mean. Is there any methods of calculating a mean with minimun variance? Without the outliers? Thanks for your help.
 

mp83

TS Contributor
#2
There are a lot of suggested measures. You can google about "robust mean statistics". The simplest one you can use is the trimmed mean.
 
#3
If you're looking for a measure of central tendency and your distribution is skewed by outliers then you should report the median. If your outliers are due to known measurement errors (ie, the measurements are incorrect) then you probably want to consider removing the incorrect values.