This is a probability bound analysis library for Python
Intervals can be specified by using pba.I(x,y)
Probability distributions can be specified using
pba.distname(**args) for all distribution that scipy.stats supports.
Using interval arguments return p-boxes
K out of N confidence boxes can be specified using pba.KN(k,n)
+,-,*,/ operations are supported. By default frechet convolutions
are used. But independant, perfect and opposite convolutions are also
supported, they can be specified using a letter as in:
A.add(B, method = 'o') # A + B using opposite convolutions C.sub(D, method = 'p') # C - D using perfect convolutions E.mul(F, method = 'i') # E * F using independence convolutions G.div(H, method = 'f') # G / H using frechet convolutions
Note: currently there may be errors in creating p-boxes for certain distribution types because of the way arguments are passed to the distributions in scipy.stats library. If these errors are noticed please email me