And now I'm sat here wondering why this isn't a default feature on calculators. It's great!
I think the simplicity really makes this work. just about anyone who can use a calculator can operate this calculator -- but that doesn't make it any less powerful. Humans are notoriously bad at intuitively dealing with uncertainty, and this calculator might very well help get a feel for it.
The only thing I'd like to see improved is the histogram, the old timey console-like graphing is cute, but not exactly easy to read or interpret. Going for a modern graphical graph and marking it with some useful percentiles would go a long way towards making it more usable.
EDIT: also -- only half facetiously -- is there a CLI/desktop/mobile app yet? Doing quick back-of-a-napkin math on your phone is a pretty useful thing.
EDIT2: Is the source available somewhere? I'm having the hardest time finding the github repo related to this page...
I already have a CLI tool and a library, but I'd like to make it a bit less hairy before releasing. I also want to build an iOS / Android app. But I wanted to release a web app first because that has by far the lowest barrier to entry.
Can someone explain the math here? How is he generating the histogram from just the mean and the range?
UPDATE
Found this:
"Range is always a normal distribution, with the lower number being two standard deviations below the mean, and the upper number two standard deviations above. Nothing fancier is possible, in terms of input probability distributions."
So the range is two std dev from the mean. He also mentions somewhere early that he's assuming the input range is 95% confident.
If I am understanding this correctly, I have a second question: if he has the mean, and the std dev, what is the purpose of the monte carlo simulation exactly? Can't you just subdivide your range, and compute frequency using the gaussian function?
This will work for simple equations, for more complex ones you will run into the curse of dimensionality. That's why MC was invented in the first place by Stanisław Ulam et al.
This is great. I often test assumptions by picking a set of low/high numbers to get a feel for the range of likelihood in a very crude way. This is more sophisticated and gives more confidence in the correctness of the answer.
I haven't bookmarked something in ages. I bookmarked this. The value added with a single new operator is amazing. Feel good about this work. Not many people can create something so unique and valuable by adding so little.
Thank you! I've been thinking about this problem for years, and had several ideas in that time. But the complexity always grew out of proportion as I started adding "features" to the notation and the tool. In the end, I'm happy with how it looks like now, even if it isn't perfect by any means.
This is great, I've done calcs to get the upper and lower bounds, but to get the full range like this is super. The example of moving was perfect, I had done a similar calculation by an excel sheet and it was a pain.
Every project you release is impressive, I gotta say. Love the UX of this one, making statistics accessible is soooo hard.
Eager to see the Flutter version!
And now I'm sat here wondering why this isn't a default feature on calculators. It's great!
I think the simplicity really makes this work. just about anyone who can use a calculator can operate this calculator -- but that doesn't make it any less powerful. Humans are notoriously bad at intuitively dealing with uncertainty, and this calculator might very well help get a feel for it.
The only thing I'd like to see improved is the histogram, the old timey console-like graphing is cute, but not exactly easy to read or interpret. Going for a modern graphical graph and marking it with some useful percentiles would go a long way towards making it more usable.
EDIT: also -- only half facetiously -- is there a CLI/desktop/mobile app yet? Doing quick back-of-a-napkin math on your phone is a pretty useful thing.
EDIT2: Is the source available somewhere? I'm having the hardest time finding the github repo related to this page...
Thank you!
I already have a CLI tool and a library, but I'd like to make it a bit less hairy before releasing. I also want to build an iOS / Android app. But I wanted to release a web app first because that has by far the lowest barrier to entry.
ah so you did think of everything ;-)
Can someone explain the math here? How is he generating the histogram from just the mean and the range?
UPDATE
Found this:
"Range is always a normal distribution, with the lower number being two standard deviations below the mean, and the upper number two standard deviations above. Nothing fancier is possible, in terms of input probability distributions."
So the range is two std dev from the mean. He also mentions somewhere early that he's assuming the input range is 95% confident.
If I am understanding this correctly, I have a second question: if he has the mean, and the std dev, what is the purpose of the monte carlo simulation exactly? Can't you just subdivide your range, and compute frequency using the gaussian function?
This will work for simple equations, for more complex ones you will run into the curse of dimensionality. That's why MC was invented in the first place by Stanisław Ulam et al.
This is great. I often test assumptions by picking a set of low/high numbers to get a feel for the range of likelihood in a very crude way. This is more sophisticated and gives more confidence in the correctness of the answer.
I haven't bookmarked something in ages. I bookmarked this. The value added with a single new operator is amazing. Feel good about this work. Not many people can create something so unique and valuable by adding so little.
Thank you! I've been thinking about this problem for years, and had several ideas in that time. But the complexity always grew out of proportion as I started adding "features" to the notation and the tool. In the end, I'm happy with how it looks like now, even if it isn't perfect by any means.
This is great, I've done calcs to get the upper and lower bounds, but to get the full range like this is super. The example of moving was perfect, I had done a similar calculation by an excel sheet and it was a pain.
Nicely done!
Maybe related: Interval Analysis ? https://en.wikipedia.org/wiki/Interval_arithmetic
Every project you release is impressive, I gotta say. Love the UX of this one, making statistics accessible is soooo hard. Eager to see the Flutter version!
Very nice project but I broke it though:
Input: 100 / 0~0
UI: Please wait...
Console: worker.dart.js:348 Uncaught Invalid argument(s): Cannot make stats from empty list of values
Fixed! Thanks for the report. This also fixes `sqrt(-10~-5)` and other fun inputs.
Amazing writeup, kudos.