Expected value (mean)

Ferra

New Member
#1
Hi All

I'm looking for a metric that shows a close trend (good correlation) to the mean (expected value) of a distribution, so I can investigate the trend of that metric instead of mean when I don't have access to mean. Any suggestion?

Thank you
 

Ferra

New Member
#3
Sorry if I didn't explain very well!

Actually, I have a number of beta distributions and I estimate summation of those as a normal dist. Instead of calculating mean of that normal dist. I want another metric that I can estimate it and that metric shows a good correlation with the mean. For example I used confidence interval, entropy,... but they didn't show good correlation...
 

BGM

TS Contributor
#6
Maybe OP wants to estimate the population mean but cannot use the sample mean. There are many different type of estimators - e.g. in other post we have talk about the sample median when the distribution is symmetric. Surely the other type of estimators is not as efficient as the sample mean but may have other types of merit, say more robust.
 

Dason

Ambassador to the humans
#7
If the beta distributions are known then why not just use the average of the expected values for the beta distributions as the mean for your normal?

... I have a feeling that I still don't understand what the issue is though...