where μ is the mean of a vector, σ2 is the variance, u2 is the average of the variances of all vectors...
for example, for the vector [0 1 0 0] the result should be 0.0833 0.3333 0.0833 0
(wiener transformed vector)
1
-
I see equations that look well-defined with constants (presumably you have multiple values of n1 and n2 and the equation applies to all values of them?), but in your explanation of the constants, it's not clear what you mean by a "vector" (in terms of the X(n1,n2)); is a vector where you fix a value of n1 and then the vector is the vector of values in X you get when you iterate over all choices of n2?user2566092– user25660922013-07-27 14:37:44 +00:00Commented Jul 27, 2013 at 14:37
Add a comment
|
1 Answer
Before implementing your own statistical stuff, have you looked into SciPy?
SciPy Stats has many modules implemented using NumPy (C written, so it's faster than Python, but callable from Python, with a Pythonic syntax). So maybe you don't need to write any code yourself.
1 Comment
Dr. Jan-Philip Gehrcke
I entirely agree and want to strengthen this point: only implement this yourself as an exercise. If you need to use this in production code, then I seriously urge you to not re-invent the wheel. Your implementation will probably be less stable, slower, and more difficult to maintain. If everybody is using the same code, this code will ultimately be the optimum (not always true, but generally a good assumption).