Random 'Normal Distribution' Numbers in Excel

[Steve Prevette]

This does provide a bit of a quandary. If you "MUST" use the normal, and the standard deviation is large enough versus the mean to provide a reasonable chance of providing negative numbers, then you are basically trapped. You are being caught between the real world situation (no negatives) and a dictated model (normal) that allows negatives. You could strip off the negative numbers generated, but then the data would not be normal.

Why MUST you use normal? What is the process you are modeling? Is it a counting process where the Poisson would be the appropriate model? Keep in mind that SPC does NOT depend upon Normality to work.

 

[Darius]

Just to mention I have to use normal distribution, not log normal.

Why is that?, can you elaborate?, as Steve explained SPC works even when non gaussian distributions (but beware after instability rules).

Random generated numbers are positive and negative numbers. Now what?

It looks like you don't have a gaussian distribution and you MUST take it as one?

If you post some data we can try to show you our way to obtain random values.

 

[bobdoering]

You could strip off the negative numbers generated, but then the data would not be normal.

Or...take your data and add the max negative number to it - then it will move the entire distribution to all positive.

Keep in mind that SPC does NOT depend upon Normality to work.

True...but it may dictate which SPC tool is correct to use.

 

[susuru]

Thank you very much to all members helping me in my "struggle" with statistics.

 

[flavorflav]

I have a similar problem... if someone can help, I would really appreciate.

We took some field observations of train dwell times, and would like to use the stats to generate new (random) dwell times based on normal distribution (well, as long as the new numbers have the same mean and standard deviation as the base data, that's fine).

For example, one location has a mean of 55 seconds and a standard deviation of 1m05s. I also want to respect the min and max values (0:22 and 3:48)

I tried using the random number generator in the Excel 2007 data analysis add on, but it only required a mean and st. dev, and no min or max. so it gives me lots of negative numbers (and numbers outside of my range)

Thanks!

 

[Jim Wynne]

I have a similar problem... if someone can help, I would really appreciate.

We took some field observations of train dwell times, and would like to use the stats to generate new (random) dwell times based on normal distribution (well, as long as the new numbers have the same mean and standard deviation as the base data, that's fine).

For example, one location has a mean of 55 seconds and a standard deviation of 1m05s. I also want to respect the min and max values (0:22 and 3:48)

I tried using the random number generator in the Excel 2007 data analysis add on, but it only required a mean and st. dev, and no min or max. so it gives me lots of negative numbers (and numbers outside of my range)

Thanks!

Instead of using the RAND function to produce your random numbers, use RANDBETWEEN which will produce numbers in the range you specify.

 

[Darius]

Can you post some data?,
the mean arround 50's, how big could be the stdev to give negative numbers?,
read this tread carefully, it will lead you to "GAUSSIAN DISTRIBUTION RANDOM GENERATION",
which function are you talking about?, the RND()*(MAX-MIN)+MIN or RANDBETWEEN funtions generate as an UNIFORM DISTRIBUTION, is that what you need?

=ROUND(22+(228-22)*RAND(),0)
=+RANDBETWEEN(22,228)

in seconds