Estimate the Mean for a Truncated Normal Distribution

[belligio]

How do you estimate the mean for a truncated normal distribution?

For example, I have a normal distribution with a mean of 20 and a standard deviation of 10, and I want to determine the average for the values of 35 or greater.

Thanks,
Tyler

[arlu1201]

Is this solved?

If you solve a problem yourself before anyone else has responded, please take a moment to describe your solution, chances are some other member will benefit.

[belligio]

Sorry, I'm still getting used to the social norms of the forum! Here's how I solved it:

First, I filled a few thousand cells with a formula that produced a random number from the normal distribution [=NORMINV(RAND(),20,10)].
Then, to calculate the average for the portion to the right of the truncation point, I just calculated the average of all the cells greater than the truncation value [=AVERAGEIF(range,<truncation point)].

I took it one step further and calculated the average for a bunch of truncation points, and I created a LINEST function of the trend. That way, I could just plug in any truncation value into the LINEST formula instead of using the huge range of randomly generated numbers, which really slows things down.

[shg]

No need for random numbers -- you can do it in closed form. See, for example, http://en.wikipedia.org/wiki/Truncat...l_distribution

Please Login or Register  to view this content.