How would you model this function /sci/?
f(0)=0
f(100)=0
f(10)=1250
I'm trying to find an answer but I have no idea how to model this function.
looks kinda like
f(x) = Ax e^(-Bx)
screw around with different values of A and B until it looks about right
>>9081556
probably a gamma distribution
>>9081556
>>9081556
You have three points; you can fit them to a degree-2 polynomial.
Your points are as follows:
(0,0)
(10,1250)
(100,0)
Explicitly, you can just model your desired polynomial in a linear form to obtain a closed form:
f(x) = a + bx + cx^2
With your three points, this yields
0 = a + 0 + 0
1250 = a + 10b + 100c
0 = a + 100b + 10000c
Which is a fully determined linear system. Solve for a,b,c and c, plug them in as your constants in f(x), and have fun with whatever you're doing.
>>9081576
You can also iteratively improve your fit by sampling more points from the curve in your image and increasing the degree of your polynomial.
>>9081556
Sample more, and see if it fits a chi square distribution.
>>9081580
You mean MLE?
>>9081576
Sorry, no, this doesn't work. (10,1250) is supposed to be the maximum, but this goes way beyond that. Sure I could do a Taylor but fuck that shit, I want to find what this distribution is without having to use such dirty tricks.
>>9081592
Doesn't look like chi square, but more like log normal.
>>9081600
No idea
https://en.wikipedia.org/wiki/Log-normal_distribution
>>9081606
I mean, if you want a quick and dirty method, you could just unlog it, then it's easy because it should be a normal distribution.
But statisticians would wag their fingers.
>>9081556
It looks like the absolute value of a potential function.