Sign up FAST! Login

Uncertainty with Ease

normal distribution

I spent yesterday learning about some newer tools to handle uncertainty, even when one has limited knowledge of statistics.  This tool -- -- lets end users do it.  You make a relatively normal spreadsheet (adding just a selection of distribution type and max/min values for any uncertain values).  Then it automatically runs Monte Carlo simulations so that everything downstream of a distribution variable is a distribution.  The results are far superior to the min/max/avg analysis that you probably do a lot, if you're like me.  Here is a blog post about the tool --

This led me to Microsoft Research's Uncertain<T>, which is way cooler if you can program, and requires only a bit more background.  I am not sure what method is used to compose distributions internally, but it seems more sophisticated.  Many operators can be handled.  What is very cool is that many programs that work with scalar (simple) values can be modified to work with uncertain distributions with only minimal changes.  Here is a white paper --

Stashed in: Math!, math, Medium

To save this post, select a stash from drop-down menu or type in a new one:

Thanks Three Pipe.

It's challenging for me to get my head around quantifying uncertainty. 

But I appreciate those who can do it. 

Joyce broke this out into its own page.

You May Also Like: