fix: use the distribution offset as min for iti statistics

[arjunsridhar12345]

Attempts to close #82 . Only uses the scaling parameters offset as the min for the iti statistics. The others remain unchanged if I understood correctly

[micahwoodard]

Should probably have a check here to take into account the min and offset. Maybe something like

    tp = distribution.truncation_parameters
    base_min = float(tp.min)

    sp = distribution.scaling_parameters
    scale = float(sp.scale) if sp is not None else 1.0
    offset = float(sp.offset) if sp is not None else 0.0

    return base_min * scale + offset

[arjunsridhar12345]

suggesting something like this?

beta: t.Optional[float] = None
params = distribution.distribution_parameters
truncation = distribution.truncation_parameters
scaling = distribution.scaling_parameters
scale = scaling.scale if scaling is not None else 1.0
offset = scaling.offset if scaling is not None else 0.0
minimum = truncation.min if truncation is not None else None
maximum = truncation.max if truncation is not None else None
if params.family == DistributionFamily.EXPONENTIAL and params.rate:
    beta = scale / params.rate
    # An exponential's support starts at the offset once shifted.
    minimum = offset if minimum is None else max(minimum, offset)
elif params.family == DistributionFamily.UNIFORM:
    # A uniform distribution carries its bounds in the distribution
    # parameters rather than the truncation parameters.
    minimum = params.min * scale + offset
    maximum = params.max * scale + offset
return beta, minimum, maximum

[micahwoodard]

Yeah I think something like that should work. Would probably be worth adding a test or something to iterate through different cases and validate

[arjunsridhar12345]

I forgot to commit this. should be here: 17b6141

[micahwoodard]

Looks good. Thanks for additions and sorry for the approval wait :)